Pulse excitation optimization control method for semiconductor field effect gas sensing array

Abstract

The invention belongs to the technical field of sensing, and relates to a pulse excitation optimization control method for a semiconductor field effect gas sensing array. A gas sensing array to which the pulse excitation optimization method faces is composed of a plurality of field effect type unit sensing devices based on semiconductor sensitive materials, and the sensitive materials of the unit sensing devices are different. According to the pulse excitation optimization method, corresponding special pulse type voltage bias is optimally designed according to the unique performance of each sensitive material, and the field effect is utilized to regulate and control the gas sensitive characteristics of each sensitive material; by further optimizing the pulse waveform, the output difference among the unit sensors is remarkably increased, the information amount of output data of the sensing array is increased, meanwhile, the array power consumption is remarkably reduced, and the effectiveness of the sensing array in the process of analyzing complex gas is improved; when applied to detection of different mixed gases, through targeted optimization of pulse excitation waveforms, the adaptive capacity of the sensing array is improved, and the research, development and application cost and difficulty of the array are reduced.

Description

Technical Field

[0001] This invention belongs to the field of sensing technology, specifically a pulse excitation optimization control method for semiconductor field-effect gas sensing arrays. Background Technology

[0002] When detecting complex gas mixtures, the accuracy and effectiveness of sensor arrays composed of multiple sensors in quantitative analysis of gas composition and concentration heavily depend on the amount of information output by the array. Increasing the amount of information primarily relies on increasing the types and number of sensor units and the differences in the output information between each unit's sensors. To meet these requirements, sensor arrays can be composed of unit devices with different signal conversion principles, such as a combination of surface acoustic wave gas sensors, quartz crystal microbalance sensors, and semiconductor resistive sensors; or they can be composed of unit devices with the same signal conversion principle but different sensitive materials, such as a combination of various metal oxide semiconductors and organic semiconductor resistive sensors. Based on these various types of commercially available discrete sensor devices, integrated onto a PCB board, a traditional sensor array or electronic nose is formed, which can be used for the quantitative identification of gas mixture composition and concentration. With the development of artificial intelligence, micro-nano fabrication technology, and MEMS technology, the construction of the Internet of Things and smart cities is progressing rapidly, placing new demands on the miniaturization and integration of sensor arrays. Integrating multiple sensors onto a single silicon wafer using fabrication methods compatible with silicon-based processes, and further improving integration density and reducing power consumption, has become an important development trend for sensor arrays. Devices based on semiconductor materials, especially field-effect semiconductor sensors, offer more flexible device control and are better suited for integration onto silicon wafers to form intelligent sensing systems, thus attracting widespread attention and research in recent years.

[0003] Existing field-effect gas sensing arrays based on semiconductor sensing materials can be composed of unit devices that use semiconductor sensing materials as gates to convert gas information into changes in the work function of the sensing material, or they can use semiconductor sensing materials as sensing channels to convert gas information into changes in resistance. Regardless of the type of semiconductor field-effect sensing array, the amount of gas information contained in its output signal mainly depends on the use of different semiconductor sensing materials to fabricate unit devices and construct the array. Once the sensing material is deposited on the substrate and the array is fabricated, under typical DC excitation control methods, the output of each unit device in the array mainly depends on the intrinsic properties of the sensing material, such as morphology, surface activity, carrier concentration, and mobility. The relatively fixed gas-sensing performance and low adjustability of these unit devices lead to low energy efficiency, high optimization difficulty, and high R&D costs for the sensor array. First, the insufficient output differences between each sensing unit device and the insufficient sensing differences of each sensing unit for different gas components in the mixed gas result in high difficulty in subsequent processing of the sensor array output signal (including feature selection and extraction), which in turn increases the complexity and power consumption of the information processing circuit in the sensing system. Second, when the detected target changes, the sensor array has poor adaptability due to its fixed configuration, control form, and performance, resulting in poor detection effect. Improving and enhancing the performance of the sensor array requires reselecting and combining the unit devices in the array, or changing the preparation process of each material in the unit devices during array fabrication, thereby adjusting the output of the sensor array, resulting in high R&D and application costs and high optimization difficulty.

[0004] In summary, the key to improving sensor array performance and reducing array development and application costs lies in how to flexibly and efficiently control the performance output of each unit sensor after the sensor array is formed, further increase the output differences of each unit sensor, and thus significantly increase the information content of the sensor array output data, thereby improving the effectiveness and accuracy of the sensor array in analyzing complex gases. Summary of the Invention

[0005] To address the aforementioned shortcomings of existing technologies, the purpose of this invention is to improve the adaptability, effectiveness, and accuracy of the sensor array in analyzing various complex gases by applying pulsed voltage excitation and optimizing the control method to increase the output differences between unit devices in a field-effect semiconductor gas sensor array that has already been fabricated, thereby significantly increasing the amount of information in the sensor array's output data, reducing research and development and application costs.

[0006] To achieve the above objectives, the present invention provides a pulse excitation optimization control method for semiconductor field-effect gas sensing arrays, the technical solution of which is as follows:

[0007] A pulse excitation optimization control method for semiconductor field-effect gas sensing arrays includes the following steps:

[0008] Step 1: Apply pulse voltage excitation to each unit device;

[0009] Step 2: Adjust the timing parameters of the pulse excitation waveform;

[0010] Step 3: Adjust the pre-bias region level parameters of the pulse excitation waveform to establish the output feature set;

[0011] Step 4: Construct a gas composition and concentration prediction model;

[0012] Step 5: Determine the optimal control method for pulse excitation.

[0013] As a further description of the above scheme, in step 1: the pulse voltage excitation waveform used includes a pre-bias region and a readout region; the initialized pulse voltage excitation is applied to each unit sensor device; and the known components X and concentration C are... X The gas sample is detected, and an electrical signal is output.

[0014] In step 2: the timing parameters of the pulse excitation waveform are optimized by determining whether the electrical signal output by the array meets the stability requirements. When the stability requirements are not met, the timing parameters of the pulse voltage excitation waveform, i.e., the pre-bias region time t, are adjusted. p and read time zone t r Make adjustments;

[0015] In step 3: when the output electrical signal meets the stability requirements, the pre-biasing time t of the pulse voltage excitation waveform is determined. p and read time t r Adjust the bias voltage level V during the pre-biasing process. p This forms a gas-sensitive control electric field, extracts features from the output information, forms a feature set, and divides it into a training set and a validation set.

[0016] In step 4: using the training set and the known composition X and concentration C of the reference gas. X Information modeling is used to obtain a prediction model M, and the predicted gas composition X′ and concentration C are validated using a validation set. X When the error between the prediction result and the validation set data does not meet the error requirements, repeat steps 2 to 3 to readjust the pulse voltage excitation waveform time parameters and pre-bias level of each unit device, and perform pre-bias and time parameter coupling optimization.

[0017] In step 5: after pre-biasing and time parameter coupling optimization of pulse voltage excitation, when the error between the prediction result and the validation set data meets the error requirements, t is determined. pt r and V p As a result, a quantitative analysis model M for the gas composition and concentration of a sensing array composed of semiconductor field-effect gas sensing unit devices with different sensitive materials was established, and a pulse excitation optimization control method for semiconductor field-effect gas sensing arrays was formed.

[0018] As a further description of the above scheme, in step 1, the pulse voltage excitation waveform applied to each unit device of the sensing array can be of one or more forms, and each pulse cycle of the pulse voltage excitation waveform includes a pre-bias region and a readout region.

[0019] As a further description of the above scheme, within the pre-bias region of pulse voltage excitation, an electric field is formed in each unit device by setting a pre-bias voltage, and no electrical signal of the device is read within this range. The pre-bias voltage of each unit device is determined after optimization based on the accuracy of the prediction results of the sensor array on the composition and concentration of the gas sample. The formed electric field will affect the adsorption and desorption between each semiconductor sensitive material and the gas, increasing the output difference of each unit device when detecting the same gas sample.

[0020] In step 1, the device electrical signal is not read within the pre-bias region of pulse voltage excitation;

[0021] In step 3, an electric field is formed in each unit device by setting a pre-bias voltage. The pre-bias voltage of each unit device is determined after optimization based on the accuracy of the prediction results of the sensor array on the composition and concentration of the gas sample. The formed electric field will affect the adsorption and desorption between each semiconductor sensitive material and the gas, increasing the output difference of each unit device when detecting the same gas sample.

[0022] As a further description of the above scheme, in step 2, within the reading area excited by pulse voltage, the sensing signals of each unit sensor are read by setting the corresponding voltage. The duration of the reading area is less than 500 microseconds, and the duration of the pre-biasing area is greater than the duration of the reading area.

[0023] As a further description of the above scheme, in step 3, feature extraction is to extract n features carrying gas information from the sensing data of each sample of the target gas, forming a response feature matrix R of k samples, which together with the known gas information set constitutes a modeling database.

[0024] As a further description of the above scheme, in step 4, a prediction model M is established using PLS regression analysis, and the formula is:

[0025] Where C lLet Rj be the concentration of the l-th component of the gas mixture G, Rj be the j-th response characteristic value, and α be the concentration of the l-th component of the gas mixture G. l =α l0 ,α l1 ,α l2 ,…,α ln [] represents the PLS regression coefficient vector corresponding to the l-th component in the gas mixture G;

[0026] The time parameter t of the pulse voltage excitation waveform of each unit device p t r and pre-bias level V p Independent settings are available to accommodate the different gas-sensing properties of various sensitive materials.

[0027] As a further description of the above scheme, the source of each unit device in the sensing array is grounded, the drain and gate are excited by the same frequency pulse voltage, the drain of each unit device in the sensing array is grounded in the pre-bias region, and a pre-bias level is applied to the gate.

[0028] The present invention has the following advantages and beneficial effects:

[0029] 1. The pulse excitation optimization control method for semiconductor field-effect gas sensor array proposed in this invention applies pulse excitation to each unit sensor in the array, performs specific gas sensing regulation on each semiconductor sensitive layer, flexibly and efficiently regulates the performance of each unit sensor, further increases the output difference of each unit sensor, and thus significantly increases the information content of the sensor array output data.

[0030] 2. The method for constructing a prediction model for the composition and concentration of mixed gases proposed in this invention can reconstruct the sensor output by re-optimizing the pulse waveform when the target mixed gas detected by the sensor array is changed, without changing the composition of the sensor array. It is highly adaptable and easy to transfer to the detection application of different mixed gases, reducing the research and development, application cost and optimization difficulty of the array. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of a pulse excitation optimization control method for semiconductor field-effect gas sensing arrays;

[0032] Figure 2 This is a schematic diagram of the pulse excitation method applied when the present invention is implemented in a 1×3 field-effect gas sensing array with tin oxide as the sensitive gate;

[0033] Figure 3 A detailed flowchart illustrating an example implementation of this invention;

[0034] Figure 4 This is a schematic diagram illustrating the information stored in the modeling database. Detailed Implementation

[0035] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the following description, using a 1×3 field-effect gas sensing array with tin oxide as the sensitive gate for qualitative and quantitative analysis of a mixed gas as an example, in conjunction with the accompanying drawings and specific embodiments, provides a detailed explanation of the specific implementation method of the present invention. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0037] A pulse excitation optimization control method for semiconductor field-effect gas sensing arrays includes the following steps:

[0038] Step 1: Apply pulse voltage excitation to each unit device;

[0039] Step 2: Adjust the timing parameters of the pulse excitation waveform;

[0040] Step 3: Adjust the pre-bias region level parameters of the pulse excitation waveform to establish the output feature set;

[0041] Step 4: Construct a gas composition and concentration prediction model;

[0042] Step 5: Determine the optimal control method for pulse excitation.

[0043] The pulse excitation optimization control method for semiconductor field-effect gas sensor arrays proposed in this invention applies pulse excitation to each unit sensor in the array, performs specific gas sensing regulation on each semiconductor sensitive layer, flexibly and efficiently regulates the performance of each unit sensor, further increases the output difference of each unit sensor, and thus significantly increases the information content of the sensor array output data.

[0044] Step 1 of the present invention: Set the initial waveform, apply pulse voltage excitation to each unit sensor, and test the known components X and concentration C. X The gas sample is detected, and an electrical signal is output.

[0045] Step 2: Determine whether the output electrical signal of the array meets the stability requirements. If the stability requirements are not met, adjust the time parameter of the pulse voltage excitation waveform, i.e., the pre-bias region time t. pand read time zone t r Make adjustments;

[0046] Step 3: When the output electrical signal meets the stability requirements, determine the pre-biasing time t of the pulse voltage excitation waveform. p and read time t r Adjust the bias voltage level V during the pre-biasing process. p This forms a gas-sensitive control electric field, extracts features from the output information, forms a feature set, and divides it into a training set and a validation set.

[0047] Step 4: Using the training set and the known composition X and concentration C of the reference gas. X Information modeling is used to obtain a prediction model M, and the predicted gas composition X′ and concentration C are validated using a validation set. X When the error between the prediction result and the validation set data does not meet the error requirements, repeat steps 2 to 3 to readjust the pulse voltage excitation waveform time parameters and pre-bias level of each unit device, and perform pre-bias and time parameter coupling optimization.

[0048] Step 5: After pre-biasing and time parameter coupling optimization of pulse voltage excitation, when the error between the prediction result and the validation set data meets the error requirements, determine t. p t r and V p As a result, a quantitative analysis model M for the gas composition and concentration of a sensing array composed of semiconductor field-effect gas sensing unit devices with different sensitive materials was established, forming a pulse excitation optimization control method for semiconductor field-effect gas sensing arrays. The method for constructing a mixed gas composition and concentration prediction model proposed in this invention can reconstruct the sensor output by re-optimizing the pulse waveform when the target mixed gas detected by the sensing array is changed, without needing to change the composition of the sensing array. This method is highly adaptable, easy to transfer to the detection of different mixed gases, and reduces the research and development, application costs, and optimization difficulty of the array.

[0049] In step 1 of the present invention, the pulse voltage excitation waveform applied to each unit device of the sensing array can be of one or more forms, and each pulse cycle includes a pre-bias region and a readout region.

[0050] In this invention, an electric field is formed in each unit device by setting a pre-bias voltage within the pre-bias region of pulse voltage excitation. Within this region, no electrical signal of the device is read. The pre-bias voltage of each unit device is determined after optimization based on the accuracy of the prediction results of the sensor array on the composition and concentration of the gas sample. The formed electric field will affect the adsorption and desorption between each semiconductor sensitive material and the gas, increasing the output difference of each unit device when detecting the same gas sample.

[0051] In step 2, the device electrical signal is not read within the pre-bias region of pulse voltage excitation;

[0052] In step 3, an electric field is formed in each unit device by setting a pre-bias voltage. The pre-bias voltage of each unit device is determined after optimization based on the accuracy of the sensor array's prediction results of the composition and concentration of the gas sample. The formed electric field will affect the adsorption and desorption between each semiconductor sensitive material and the gas, increasing the output difference of each unit device when detecting the same gas sample.

[0053] In step 3 of the present invention, within the reading area excited by pulse voltage, the sensing signals of each unit sensor are read by setting a corresponding voltage to form an array output set. The duration of the reading area is less than 500 microseconds, and the duration of the pre-bias area is greater than the duration of the reading area.

[0054] In step 3 of this invention, feature extraction involves extracting n features carrying gas information from the sensing data of each sample of the target gas, forming a response feature matrix R of k samples, and combining it with the known gas information set to form a modeling database.

[0055] In step 4 of this invention, a prediction model M is established using PLS regression analysis, and the formula is as follows:

[0056] Where C l R is the concentration of the l-th component of the gas mixture G. j Let α be the j-th response feature value. l =[α l0 ,α l1 ,α l2 ,…,α ln [] represents the PLS regression coefficient vector corresponding to the l-th component in the gas mixture G;

[0057] The time parameter t of the pulse voltage excitation waveform of each unit device p t r and pre-bias level V p Independent settings are available to accommodate the different gas-sensing properties of various sensitive materials.

[0058] In this invention, the source of each unit device in the sensing array is grounded, and the drain and gate are excited by a pulse voltage of the same frequency. In the pre-bias region, the drain of each unit device in the sensing array is grounded, and a pre-bias level is applied to the gate.

[0059] Example:

[0060] This example illustrates a specific implementation method for high-precision mixed gas G detection using the pulse excitation optimization control method of this invention on a field-effect 1×3 tin oxide sensitive grid gas sensing array.

[0061] There are k samples of the known gas mixture G being tested, and each sample contains m components. The corresponding component information set is X = [x1, x2, ..., x...]. l ,…x m ] T The concentrations of the m components in the i-th sample are C. iX =[C i1 C i2 ,…,C il ,…C im If the gas information set of k samples is C, then C is the gas information set of k samples. X =[C 1X C 2X ,…,C iX ,…,C kX ] T .

[0062] like Figure 2 As shown, a pulse excitation is applied to a 1×3 field-effect gas sensor array with tin oxide as the sensitive gate. The three sensor units in the array are S1, S2, and S3. The sources of all three sensor units are grounded, and their drains and gates are excited by a pulsed voltage at the same frequency. One repetition cycle of the voltage waveform includes a pre-bias region and a readout region. The readout region durations of the three sensor units are t0, t1, t2, t3 ... r1 t r2 t r3 The pre-bias region durations are t p1 t p2 t p3 During the readout period, a drain readout level of V is applied to the drain of each of the three sensing units. dr1 V dr2 V dr3 The gates are respectively applied with gate readout level V. gr1 V gr2 V gr3 This is used to read the output signals of each sensor unit. During the pre-bias region, the drains of all three sensing units are grounded, and a gate pre-bias level, V, is applied to the gate. p1 V p2 V p3 The three pre-bias levels will generate different degrees of field effect within the tin oxide sensitive layer of the three unit sensor devices, thereby modulating the degree of reaction between the sensitive channel layer and the gas.

[0063] like Figure 1-3 As shown, the specific implementation steps of the pulse excitation optimization control method for the sensor array in this embodiment are as follows:

[0064] (1) Targeting Figure 2The three sensors S1, S2, and S3 in the sensor array are respectively set to the drain readout level V of the same frequency pulsed voltage excitation waveform at the drain and gate. dr1 V dr2 V dr3 and gate read level V gr1 V gr2 V gr3 ;

[0065] (2) Targeting Figure 2 The three sensors S1, S2, and S3 in the sensing array are each configured with a pre-bias level V on their gates. p1 V p2 V p3 ;

[0066] (3) Targeting Figure 2 The three sensors in the sensing array are initialized with time parameters of the same-frequency pulsed voltage excitation waveforms at the drain and gate, including the duration t of the three readout regions. r1 t r2 t r3 And the duration t of the three pre-bias regions p1 t p2 t p3 ;

[0067] (4) Apply the set pulse excitation to the sensor array and test the i-th sample among the k samples of the mixed gas G with known composition and concentration, and collect instantaneous sensing data.

[0068] (5) Calculate the noise and signal-to-noise ratio of the collected instantaneous sensing data to analyze whether it meets the signal stability requirements. If it does not meet the requirements, repeat (3) to adjust the pulse excitation waveform time parameters, and repeat (4) and (5). If it meets the signal stability requirements, continue to the next step.

[0069] (6) Adjust the pre-bias level V of each unit device. p1 V p2 V p3 The sensor data is output by sequentially detecting and testing k samples of gas G with known composition and concentration.

[0070] (7) Extract n features carrying gas information from the sensing data of each sample of gas G to form a response feature matrix R = [R1, R2, ..., R] for k samples. i ,…,R k ] T The response feature vector of the i-th sample is R. i =[R i1 ,R i2 ,…,R ij,…,R in [, and together with the composition and concentration information of k samples of known gas G, form a gas information set C.] X =[C 1X C 2X ,…,C iX ,…C kX ] T Together they form the modeling database, and this process is as follows: Figure 4 As shown;

[0071] (8) Divide the modeling database into a test set and a training set, using R from the training set as input data, C X As output data, a predictive model M is built using PLS regression analysis:

[0072]

[0073] Where C l R is the concentration of the l-th component of the gas mixture G. j Let α be the j-th response feature value. l =[α l0 ,α l1 ,α l2 ,…,α ln [α] represents the PLS regression coefficient vector corresponding to the l-th component in the gas mixture G. The PLS regression coefficient matrix for all gas components is α = [α1, α2, ..., α]. m ] T ;

[0074] (9) Take R from the test set as input data, input it into the prediction model M, and output the predicted components and concentration dataset C. X ′, and perform error analysis based on the true values ​​in the test set. If the error does not meet the requirements, return (2). If the error meets the requirements, determine the parameters of the pulse excitation waveform and output the prediction model M.

[0075] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the present invention is not limited to the above embodiments and there can be many variations. Any non-substantial improvements made using the inventive concept and technical solution of the present invention, or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.

Claims

1. A method for pulse excitation optimization control of a semiconductor field effect gas sensing array, characterized in that, The method comprises the following steps: Step 1: applying pulse voltage excitation to each unit device; Step 2: adjusting the time parameters of the pulse excitation waveform; Step 3: adjusting the pre-bias region level parameters of the pulse excitation waveform to establish an output feature set; Step 4: constructing a gas composition and concentration prediction model; Step 5: determining the pulse excitation optimization control method.

2. The pulse excitation optimization control method for a semiconductor field effect gas sensing array according to claim 1, characterized in that: In the step 1: the pulse voltage excitation waveform adopted contains pre-bias region and reading region, the initialized pulse voltage excitation is applied to each unit sensor device, and the known composition component X and concentration C X The gas sample is detected, and an electric signal is output. In the step 2, the time parameters of the pulse excitation waveform are optimized by judging whether the electrical signal outputted by the array meets the stability requirement, and when the stability requirement is not met, the time parameters of the pulse voltage excitation waveform, i.e. the pre-bias region time t p and the reading time region t r are adjusted; In step 3, when the output electrical signal meets the stability requirement, the pre-bias region time t of the pulse voltage excitation waveform is determined p And the reading region time t r The bias voltage level V in the pre-bias process is adjusted p A gas-sensitive regulation electric field is formed, the output information is feature extracted to form a feature set, and the feature set is divided into a training set and a verification set; In step 4, the training set and the known reference gas composition X and concentration C are used X Information modeling to obtain a prediction model M, and using the validation set to verify the predicted gas composition X' and concentration C X When the error between the prediction result and the validation set data does not meet the error requirement, steps 2 to 3 are repeated, the pulse voltage excitation waveform time parameters and the pre-bias level of each unit device are re-adjusted, and the pre-bias and time parameter coupling optimization is performed; In step 5, after the pre-bias and time parameter coupling optimization under the impulse voltage excitation, when the error between the prediction result and the validation set data meets the error requirement, t p , t r and V p are determined, a sensing array gas component and concentration quantitative analysis model M is established by the semiconductor field effect gas sensing unit devices with different sensitive materials, and a pulse excitation optimization control method for the semiconductor field effect gas sensing array is formed.

3. The method of claim 2, wherein the method is used for a pulsed excitation optimized control of a semiconductor field effect gas sensing array. In step 1, the pulse voltage excitation waveform applied to each unit device of the sensing array can have one or more forms, and each pulse cycle of each form of the pulse voltage excitation waveform includes a pre-bias region and a reading region.

4. The method of claim 2, wherein the method is characterized by: In the pre-bias region of the pulse voltage excitation, a pre-bias voltage is set to form an electric field in each unit device, and no device electrical signal is read in this region. The pre-bias voltage applied to each unit device is determined after optimization according to the accuracy of the prediction results of the sensing array on the composition and concentration of the detected gas sample. The formed electric field will affect the adsorption and desorption between each semiconductor sensitive material and the gas, increasing the output difference of each unit device when detecting the same gas sample. In step 1, no device electrical signal is read in the pre-bias region of the pulse voltage excitation. In step 3, a pre-bias voltage is set to form an electric field in each unit device. The pre-bias voltage applied to each unit device is determined after optimization according to the accuracy of the prediction results of the sensing array on the composition and concentration of the detected gas sample. The formed electric field will affect the adsorption and desorption between each semiconductor sensitive material and the gas, increasing the output difference of each unit device when detecting the same gas sample.

5. The method of claim 2, wherein the method is characterized by: In step 2, in the reading region of the pulse voltage excitation, the sensing signal of each unit sensor is read by setting the corresponding voltage. The duration of the reading region is less than 500 microseconds, and the duration of the pre-bias region is greater than that of the reading region.

6. The method of claim 2, wherein the method is characterized by: In step 3, feature extraction is extracting n features carrying gas information from the sensing data of each sample of the target gas to form a response feature matrix R of k samples, and together with the known gas information set to form a modeling database.

7. The method of claim 2, wherein the method is characterized by: In step 4, a prediction model M is established using PLS regression analysis, and the formula is where C l is the concentration of the lth component of the mixed gas G, R j is the jth response characteristic value, α l = [α l0 , α l1 , α l2 ,..., α ln ] is the PLS regression coefficient vector corresponding to the lth component in the mixed gas G; The time parameters tp, tr and pre-bias level Vp of the pulse voltage excitation waveform of each unit device are independently set to adapt to the gas sensing characteristics of different sensitive materials.

8. The method of claim 2, wherein the method is characterized by: The source of each unit device of the sensing array is grounded, the drain and gate are excited by the same frequency pulse voltage, and the drain of each unit device of the sensing array in the pre-bias region is grounded, and the gate is applied with a pre-bias level.

Images