Method for applying star flash to MEMS accelerometer integrated AI chip

Abstract

The invention relates to a method for applying star flash to an MEMS accelerometer set AI chip, and belongs to the field of intelligent manufacturing, and the method comprises the steps: carrying out the source positioning of a wireless signal in a sensor environment of densely deployed MEMS accelerometers through employing a short-distance communication protocol; transmitting and receiving data by adopting a radio wave transmission mode of a long-distance communication protocol, and mapping a plurality of radio wave signal attenuation models; an AI chip is deployed on an inductor, a spiking neuron model is constructed by using the AI chip, and the spiking neuron model is used for carrying out change analysis and prediction on input and output voltage data received by the inductor. According to the probability value, the relation between the multi-pulse input synapses of a plurality of input commands requiring acceleration and deceleration at the same time and the excitation threshold value is sequenced, and command output is performed according to the high-low sequence of the predicted values so as to determine the trend relation between the output voltage and the acceleration. The method can improve the positioning precision of the sensor.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent manufacturing, and in particular relates to a method for applying StarFlash to a MEMS accelerometer integrated with an AI chip. Background Art

[0002] The Starflash Alliance was officially established on September 22, 2020. So far, there are more than 140 member units. Starflash technology can meet the needs of smart cars, smart terminals, smart manufacturing, and smart homes for high-quality short-range wireless connections.

[0003] The term MEMS (Micro-Electro-Mechanical System) is often used to describe the types of sensors and the process of making sensors. MEMS is also called micro-electro-mechanical system, microsystem, micro-machine, etc., and refers to high-tech devices with a size of a few millimeters or even smaller. The internal structure of a micro-electro-mechanical system is generally in the micron or even nanometer level, and it is an independent intelligent system. MEMS consists of mechanical microstructures, microsensors, microactuators and microelectronics, and all structures are integrated on the same silicon substrate.

[0004] MEMS sensors are new sensors made with micromachining technology and are an important branch of MEMS. There are many types of MEMS sensors, which can be divided into three categories according to their working principles: physical, chemical and biological. According to the measured value, they can be divided into sensors such as acceleration, angular velocity, pressure, displacement, flow, electric quantity, magnetic field, infrared, temperature, gas composition, humidity, pH value, ion concentration, biological concentration and touch.

[0005] With the development of MEMS technology, inertial sensor devices have become one of the most widely used micro-electromechanical system devices, and micro-accelerometers are typical representatives of inertial sensor devices. The theoretical basis of micro-accelerometers is Newton's second law of motion. The acceleration of a system can be measured. If the initial velocity is known, the linear velocity can be calculated by integration, and then the linear displacement can be calculated. Combined with a gyroscope (used to measure angular velocity), objects can be accurately positioned.

[0006] The impact of huge computing power consumption on sensor devices is mainly reflected in the following two aspects:

[0007] First, the higher the computational complexity of deep neural networks in the inference stage, the more difficult it is to meet some smart manufacturing computing scenarios that have high real-time requirements, such as autonomous driving, risk control detection, etc., which makes model performance and computing real-time the main contradiction.

[0008] Second, the huge amount of computing will also result in a lot of energy consumption. Summary of the invention

[0009] In view of the above deficiencies in the prior art, the purpose of the present invention is to provide a method for applying Star Flash to a MEMS accelerometer integrated with an AI chip, so as to improve the positioning accuracy of the sensor and enhance the application scope and efficiency of the intelligent enhancement of inertial sensor devices using MEMS technology.

[0010] The present invention proposes a method for applying Star Flash to a MEMS accelerometer integrated with an AI chip, comprising:

[0011] S1, using a short-distance communication protocol to locate the source of wireless signals in a sensor environment where MEMS accelerometers are densely deployed; using a long-distance communication protocol for radio wave transmission to send and receive data, mapping out a variety of radio wave signal attenuation models, wherein the radio wave signal attenuation model is a model of the impact of different weather conditions on the strength of terahertz signals;

[0012] S2, deploying an AI chip on the sensor, and using the AI ​​chip to build a pulse neuron model, the pulse neuron model is used to analyze and predict the changes of the input and output voltage data received by the sensor, and sort the relationship between the multi-pulse input synapses and the excitation thresholds of multiple input commands requiring acceleration and deceleration at the same time by probability value, and output the command according to the high and low ranking of the predicted values ​​to determine the trend relationship between the output voltage and the acceleration;

[0013] S21, constructing a pulse neuron model using the AI ​​chip, wherein the neuron network of the pulse neuron model includes a pulse input point, a synaptic point, an excitation function, an output point, and a reset time;

[0014] S22, using the trend between the Markov chain output voltage and acceleration to predict the probability of the next acceleration trend of the inertial sensor device, sorting the relationship between the multi-pulse input synapses and the excitation thresholds of multiple simultaneous input commands requiring acceleration and deceleration by probability value, and giving priority to outputting commands with high prediction values.

[0015] Furthermore, in S1, the short-range communication protocol adopts the IEEE802.15.3a channel model.

[0016] Furthermore, the IEEE802.15.3a channel model is expressed as:

[0017]

[0018] Where X represents the log-normal shadow attenuation; a k，l represents the attenuation coefficient of the kth path in the lth cluster; L represents the number of clusters; l represents the index of L; K represents the number of multipaths in each cluster; k represents the index of K; T l represents the arrival time of the lth cluster; τk，l represents the delay of the kth path in the lth cluster relative to the first path; t represents time, and δ() represents the time function of channel attenuation;

[0019] Among them, a k，l Defined as;

[0020] a k，l =p k，l β k，l

[0021] Among them, p k，l represents a discrete random variable with equal probability of taking +1 and -1; β k，l It indicates that the channel coefficient of the k-th path in the l-th cluster follows the log-normal distribution.

[0022] Further, in S1, the radio wave signal attenuation model includes: an attenuation value model of rain communication signals, an attenuation value model of fog communication signals, an attenuation value model of snowfall communication signals and an attenuation value model of sandstorm climate characteristics.

[0023] Furthermore, the attenuation value model of the rain communication signal is used to characterize the relationship between the wireless signal attenuation rate and rainfall, wherein the attenuation value γ of the rain communication signal is R The model is:

[0024] γ R =KR A

[0025] Among them, R is the rainfall; K and A are parameters in two different polarization conditions: horizontal and vertical.

[0026] Furthermore, the attenuation value model of the cloud-fog communication signal is used to characterize the relationship between the wireless signal attenuation rate and visibility, wherein the attenuation value γ of the cloud-fog communication signal is c&f The model is:

[0027]

[0028] Where f is the operating frequency, V m For visibility.

[0029] Furthermore, the attenuation value model of the snowfall communication signal is used to characterize the relationship between the attenuation rate of the wireless signal and the change of the snowfall intensity, wherein the attenuation value γ of the snowfall communication signal is s The model is:

[0030] γ s =7.47×10 -5 f×I(1+5.77×10 -5 f 3 I 0.6 )

[0031] Among them, f is the operating frequency, I is the snowfall intensity;

[0032] Furthermore, the dust climate characteristic attenuation value model is used to characterize the relationship between the wireless signal attenuation rate and dust, wherein the dust climate characteristic attenuation value L R The model is:

[0033]

[0034] Where ε' and ε" are the real and imaginary parts of the complex dielectric constant of wet dust, respectively, and N 0 is the volume density of naturally and artificially formed dust samples, m is the mean of lnD, σ is the standard deviation of lnD, f is the frequency, and lnD is the logarithm of the dust particle size D.

[0035] Further, in S22, the probability of the trend between the historical output voltage and acceleration, the probability of the trend between the current output voltage and acceleration changing from acceleration to deceleration, and the probability of the trend between the current output voltage and acceleration changing from deceleration to acceleration are predicted to predict the probability that the next output voltage and acceleration trend will change from acceleration to deceleration and the probability that the next output voltage and acceleration trend will change from deceleration to acceleration.

[0036] Furthermore, the MEMS accelerometer includes the following types: a piezoresistive micro-accelerometer, a capacitive micro-accelerometer and a piezoelectric micro-accelerometer.

[0037] The beneficial effects of the present invention are as follows:

[0038] (1) The use of short-range communication protocols can solve the problem of insufficient positioning accuracy of traditional sensors in an environment where sensors are densely deployed, and improve the positioning accuracy of sensors.

[0039] (2) Long-distance communication uses millimeter wave, terahertz and other radio wave transmission methods to send and receive data and reduce the impact of abnormal weather on signals, thereby increasing the application scope of the intelligent enhancement of inertial sensor devices based on MEMS technology and the acceleration speed of sensors on objects.

[0040] (3) MEMS accelerometer is a miniature accelerometer, which is an instrument used to measure the acceleration of an object. Compared with traditional accelerometers, MEMS accelerometers are smaller in size and mass and have been widely used in intelligent manufacturing equipment. The working principle of MEMS accelerometer is as follows: when an external object accelerates, the displacement of the mass block is affected, and the capacitance between the movable arm and the fixed arm (sensor) will change accordingly: if the change in the output voltage of the sensor is measured, that is, the displacement of the actuator (mass block) is measured, the displacement has a definite corresponding relationship with the acceleration to be measured, and the output voltage and acceleration also have a definite relationship.

[0041] (4) Deploy AI chips on the sensors, and build pulse neuron models. The neural network consists of five parts: pulse input points, synaptic points, excitation functions, output points, and reset time. The chip builds a neural network to analyze and predict the changes in the input and output voltage data received by the sensor, and sorts the relationship between the multi-pulse input synapses and the excitation thresholds of multiple input commands requiring acceleration and deceleration at the same time by probability values. Commands with high prediction values ​​are given priority output, thereby determining the trend relationship between the output voltage and acceleration to complete the intelligent improvement of inertial sensor devices using MEMS technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are only used to illustrate specific embodiments and are not considered to limit the present invention. In the entire drawings, the same reference symbols represent the same components. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0043] Figure 1 This is a flow chart of a method for applying Star Flash to a MEMS accelerometer integrated with an AI chip according to an embodiment of the present invention;

[0044] Figure 2 Schematic diagram of a spiking neuron model in an SNN according to an embodiment of the present invention;

[0045] Figure 3 This is a schematic diagram of the principle of a silicon piezoresistive micro-acceleration sensor according to an embodiment of the present invention;

[0046] Figure 4 A classification diagram of MEMS sensors according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work should fall within the scope of protection of the present invention.

[0048] Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts disclosed in the present invention.

[0049] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the orientation or position relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. The terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0050] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of methods and systems consistent with some aspects of the present invention as detailed in the appended claims.

[0051] The technical terms involved in the present invention are explained below:

[0052] UWB: UltraWideBand (UWB) technology is a wireless carrier communication technology that does not use a sinusoidal carrier but instead uses nanosecond-level non-sinusoidal narrow pulses to transmit data, so it occupies a wide spectrum range.

[0053] AI chips are also called AI accelerators or computing cards, which are modules specifically designed to handle large amounts of computing tasks in artificial intelligence applications (other non-computing tasks are still handled by the CPU).

[0054] like Figure 1 As shown, the present invention proposes a method for applying Star Flash to a MEMS accelerometer integrated with an AI chip, comprising:

[0055] S1 uses a short-distance communication protocol to locate the source of wireless signals in a sensor environment where MEMS accelerometers are densely deployed; uses a long-distance communication protocol for radio wave transmission to send and receive data, mapping out a variety of radio wave signal attenuation models. Among them, the radio wave signal attenuation model is a model of the impact of different weather conditions on the strength of terahertz signals.

[0056] Specifically, the use of short-distance communication protocols can improve the problem of insufficient positioning accuracy of traditional sensors in environments where sensors are densely deployed. At the same time, long-distance communication uses radio wave transmission methods such as millimeter waves and terahertz to send and receive data and reduce the impact of abnormal weather on signals. This will increase the application scope and efficiency of inertial sensor devices.

[0057] (1) Short-distance communication protocol

[0058] In this step, the short-range communication protocol adopts the IEEE802.15.3a channel model.

[0059] The use of short-range communication protocols can improve the problem of insufficient positioning accuracy of traditional sensors in an environment where sensors are densely deployed. First, the source of the wireless signal is located. At present, it is difficult for traditional positioning technology to accurately locate multiple transmitting signal sources within an area (1-3 meters). Therefore, IEEE802.15.3a is a short-range (within 10m) and high-speed data transmission (110~480Mbit / s) channel model, which is evolved from the SV model. This model retains the characteristics of multipath clustered arrival and energy obeying double exponential distribution in the SV model. At the same time, it has made certain corrections based on actual measurement results, changing the statistical characteristics of multipath attenuation from the original Rayleigh distribution to the current log-normal distribution, and changing the channel coefficient to be expressed in real numbers.

[0060] Specifically, the IEEE802.15.3a channel model is expressed as:

[0061]

[0062] Where X represents the log-normal shadow attenuation; a k，l represents the attenuation coefficient of the kth path in the lth cluster; L represents the number of clusters; l represents the index of L; K represents the number of multipaths in each cluster; k represents the index of K; T l represents the arrival time of the lth cluster; τ k，l represents the delay of the kth path in the lth cluster relative to the first path; t represents time, and δ() represents the time function of channel attenuation;

[0063] Among them, the channel coefficient a k，l Defined as;

[0064] a k，l =p k，l β k，l (2)

[0065] Among them, p k，l represents a discrete random variable with equal probability of taking +1 and -1; β k，l represents the channel coefficient of the kth path in the lth cluster that follows the log-normal distribution;

[0066] β k，l It can be expressed as

[0067]

[0068] Among them, x k，l Indicates that the mean is μ k，l The standard deviation is σ k，l is a Gaussian random variable.

[0069] The lognormal shadow model plays an important role in the design and analysis of wireless communication systems. It can be used to predict and evaluate the impact of obstacles on wireless signal propagation, thereby optimizing the performance of wireless communication systems.

[0070] IEEE802.15.3a provides reference values ​​in different environments. According to the average distance between the transmitting and receiving nodes and whether there is line of sight, four different standard measurement channels are given. Table 1 shows the four channel models corresponding to IEEE802.15.3a.

[0071] Table 1

[0072] Model Type Whether line of sight Distance / m CM1 LOS 0～4 CM2 NLOS 0～4 CM3 NLOS 4～10 CM4 NLOS Extreme

[0073] (2) Long distance communication protocol

[0074] Radio wave signal attenuation model: Different models for different natural weather conditions, mainly including four typical weather conditions (rain communication, cloud and fog communication, snow communication, and sandstorm climate).

[0075] Terahertz signals are characterized by fast transmission and long distance. However, they are easily affected by natural weather, causing the signal to weaken and fade. The present invention maps the influence model of different weather conditions on the strength of terahertz signals by fixing the power consumption value of the radio waves sent by the transmitter and the radio waves actually received by the receiver.

[0076] In an embodiment of the present invention, the radio wave signal attenuation model includes: an attenuation value model of rain communication signals, an attenuation value model of cloud communication signals, an attenuation value model of snowfall communication signals and an attenuation value model of sandstorm climate characteristics.

[0077] (1) The attenuation value model of the rain communication signal is used to characterize the relationship between the attenuation rate of the wireless signal and the amount of rainfall. When the radio wave is transmitted in line of sight, in addition to attenuation in the air, it will also encounter attenuation in rain. If it is transmitted in rain, the attenuation rate γ caused by rainfall is as shown in formula (4). The attenuation value γ of the rain communication signal is R The model is:

[0078] γ R =KR A (4)

[0079] Among them, R is the rainfall in mmh; K and A are parameters for horizontal and vertical polarization respectively.

[0080] According to formula (4) and taking different frequencies and rainfall amounts, the relationship between the attenuation rate and rainfall amount can be obtained through simulation. The relationship between the attenuation rate and the frequency and rainfall amount shows that the attenuation rate increases with the increase of rainfall, and the growth rate is faster.

[0081] (2) The attenuation value model of cloud and fog communication signals is used to characterize the relationship between the attenuation rate of wireless signals and visibility. The attenuation rate of clouds and fog can be calculated using an empirical formula. The specific algorithm is shown in Equation 5). c&f The model is:

[0082]

[0083] Where f is the operating frequency in GHz: V m For visibility.

[0084] The international regulations on visibility are: dense fog, V < 50m; thick fog, 50m < V < 200m; light fog, 200m < < 500m. The visibility range of the attenuation rate with the frequency and visibility is 20 to 500m. The analysis frequency range of the atmospheric attenuation rate of the present invention is 0 to 350GHz. Value setting unit: dB / km and GHz20 to 500m (visibility)

[0085] (3) The attenuation value model of snowfall communication signal is used to characterize the relationship between the attenuation rate of wireless signal and the change of snowfall intensity. The radio wave attenuation rate y caused by snowfall can be approximately expressed by formula (6). Among them, the attenuation value γ of snowfall communication signal is s The model is:

[0086] γ s =7.47×10 -5 f×I(1+5.77×10 -5 f 3 I 0.6 ) (6)

[0087] Where f is the operating frequency in Gz; I is the snowfall intensity (mmh), and y is in dm, which is the height of the snow melted into water per unit volume per hour. The attenuation varies with frequency and snowfall intensity. For frequencies above 150GHz, the snowfall intensity has a greater impact, while for frequencies below 150GHz, the snowfall intensity has a relatively small impact.

[0088] Value setting unit: Frequencies above 150GHz are greatly affected by the intensity of snowfall, and an alarm is recommended.

[0089] (4) The dust climate characteristic attenuation value model is used to characterize the relationship between the wireless signal attenuation rate and dust.

[0090] Value setting unit: Below 20 GHz, the characteristic attenuation rate increases significantly with increasing frequency. Above 20 GHz, the characteristic attenuation rate increases relatively slowly with increasing frequency.

[0091] Northwest China and North China are relatively dry, and often experience three types of weather conditions: floating dust, blowing sand, and sandstorms, which have a significant impact on the attenuation of communication signals. In sandstorm weather, there are large sand particles and small dust particles, which have a greater impact on the propagation of millimeter waves and terahertz waves. Dust can be divided into two types: natural and artificial. Among them, artificially formed dust involves dust generated by explosions or moving vehicles. The shape of sand particles is complex and diverse. The shape of sand particles varies depending on the environment and the causes of dust in different regions. The distribution of sand particles is close to the log-normal distribution, and its calculation method is shown in formula (7).

[0092]

[0093] Where: N is the volume density of dust (1 / m 3 ) ; D is the diameter of the dust particles; p(D) is the density function of the size distribution of the dust particles; m is the mean of lnD; σ is the standard deviation of lnD.

[0094] The calculation method of the characteristic attenuation rate L is shown in formula (8). Among them, the characteristic attenuation value of the dust climate L R The model is:

[0095]

[0096] Where ε' and ε" are the real and imaginary parts of the complex dielectric constant of wet dust, respectively, and N 0 is the volume density of naturally and artificially formed dust samples, m is the mean of lnD, σ is the standard deviation of lnD, f is the frequency, and lnD is the logarithm of the dust particle size D.

[0097] It should be noted that in the dust climate characteristic attenuation value model, m usually represents the mean value of the logarithm of the sand particle radius r, lnr, which can also be understood as the mean value of lnD (where D represents the dust particle size). This parameter plays a key role in the model and is used to describe the size distribution of dust particles.

[0098] S2, deploy AI chips on the sensor, and use the AI ​​chips to build a pulse neuron model. The pulse neuron model is used to analyze and predict the changes in the input and output voltage data received by the sensor, and sort the relationship between the multi-pulse input synapses and the excitation thresholds of multiple input commands requiring acceleration and deceleration at the same time through probability values, and output commands according to the high and low ranking of the predicted values ​​to determine the trend relationship between the output voltage and acceleration.

[0099] Specifically, a MEMS accelerometer is a miniature accelerometer, which is an instrument used to measure the acceleration of an object. Compared with traditional accelerometers, MEMS accelerometers are smaller in size and mass, and have been widely used in intelligent manufacturing equipment. Its working principle is that the acceleration of an external object affects the displacement of the mass block, and the capacitance between the movable arm and the fixed arm (sensor) will change accordingly: if the change in the output voltage of the sensor is measured, that is, the displacement of the actuator (mass block) is measured, the displacement has a definite corresponding relationship with the acceleration to be measured, and the output voltage and acceleration also have a definite relationship.

[0100] The present invention deploys an AI chip on the sensor, and the AI ​​chip constructs a pulse neuron model. The neuron network includes five parts: pulse input point, synaptic point, excitation function, output point and reset time. The chip constructs a neural network for analyzing and predicting the changes in the input and output voltage data received by the sensor, and sorts the relationship between the multi-pulse input synapse and the excitation threshold of multiple input commands requiring acceleration and deceleration at the same time by probability value, and outputs commands with high prediction values ​​first. In this way, the trend relationship between the output voltage and the acceleration is determined to complete the intelligent improvement of the inertial sensor device of MEMS technology. Figure 4 Shown is a classification of MEMS sensors.

[0101] The main MEMS products include micro pressure sensors, inertial measurement devices, micro flow systems, read / write heads, optical systems, printer nozzles, etc., among which products for the automotive industry and information industry account for about 80% of sales.

[0102] S21, using AI chips to build a spiking neuron model.

[0103] like Figure 2 As shown, the neuron network of the pulse neuron model includes pulse input points, synaptic points, excitation functions, output points and reset time.

[0104] S22, using the trend between the Markov chain output voltage and acceleration to predict the probability of the next acceleration trend of the inertial sensor device, sorting the relationship between the multi-pulse input synapses and the excitation thresholds of multiple simultaneous input commands requiring acceleration and deceleration by probability value, and giving priority to outputting commands with high prediction values.

[0105] Specifically, the probability of the trend between the historical output voltage and acceleration, the probability of the current trend between the output voltage and acceleration transferring from acceleration to deceleration, and the probability of the current trend between the output voltage and acceleration transferring from deceleration to acceleration are predicted to predict the probability that the next trend between the output voltage and acceleration will change from acceleration to deceleration, and the probability that the next trend between the output voltage and acceleration will change from deceleration to acceleration.

[0106] Markov prediction model formula: X(k+1)=X(k)×P(9)

[0107] Among them, X(k) represents the state vector of the trend analysis and prediction object at time t=k, P represents the one-step transfer probability matrix, and X(k+1) represents the state vector of the trend analysis and prediction object at time t=k+1.

[0108] Trend probability between historical output voltage and acceleration: [0.3, 0.7]

[0109] The probability of the trend between the output voltage and acceleration changing from acceleration to deceleration is: [0.6, 0.4]

[0110] The probability of the trend between the output voltage and acceleration changing from deceleration to acceleration is [0.3, 0.7]

[0111] The probability that the trend between the next output voltage and acceleration changes from acceleration to deceleration: 39%

[0112] 0.3x0.6+0.3x0.7=0.39

[0113] The probability that the trend between the next output voltage and acceleration changes from deceleration to acceleration: 61%

[0114] 0.3x0.4+7x0.7=0.61

[0115] S23, MEMS accelerometer process.

[0116] MEMS accelerometer, or miniature accelerometer, is an instrument used to measure the acceleration of an object. Compared with traditional accelerometers, MEMS accelerometers are smaller in size and mass, and have been widely used in intelligent manufacturing equipment. Its working principle is that when the accelerometer accelerates together with the external object, the mass block moves in the opposite direction due to the inertial force, and its displacement is limited by the spring and damper. When the external acceleration is constant, the mass block has a certain displacement: when the external acceleration changes, the displacement of the mass block also changes accordingly. When the displacement of the mass block changes, the capacitance between the movable arm and the fixed arm (sensor) will change accordingly: if the change in the output voltage of the sensor is measured, that is, the displacement of the actuator (mass block) is measured, the displacement has a certain corresponding relationship with the acceleration to be measured, and the output voltage and acceleration also have a certain relationship.

[0117] V m Represents the input voltage signal, V s Indicates the output voltage, C s1 With C s2 Represent the two capacitors between the fixed arm and the movable arm respectively, then the relationship between the input signal and the output signal can be expressed as:

[0118]

[0119] where the relationship between capacitance and displacement is given by the definition of capacitance:

[0120]

[0121] Where x is the displacement of the movable arm (actuator); d is the distance between the fixed arm and the cantilever when there is no acceleration; ε 0 and ε are capacitance parameters. From equations (9) and (10), we can get

[0122]

[0123] According to the principles of mechanics, the mechanical equation of the mass block under stable conditions is:

[0124] kx=-ma ext (12)

[0125] Where k is the spring constant; m is the mass of the mass block.

[0126] Therefore, the relationship between external acceleration and output voltage is:

[0127]

[0128] It can be seen from formula (13) that when the structure and input voltage of the accelerometer are determined, the output voltage is proportional to the acceleration.

[0129] MEMS accelerometers can be divided into piezoresistive microaccelerometers, capacitive microaccelerometers, piezoelectric microaccelerometers, etc. according to the measurement principle.

[0130] In an embodiment of the present invention, the MEMS accelerometer includes the following types: a piezoresistive micro-accelerometer, a capacitive micro-accelerometer, and a piezoelectric micro-accelerometer.

[0131] (1) Piezoresistive micro-accelerometer

[0132] When a certain stress is applied to a certain axis of a semiconductor, the phenomenon that its resistivity changes is called the piezoresistive effect of the semiconductor. The principle of the piezoresistive micro-accelerometer is that when there is an external acceleration input, the mass block will be affected by an inertial force, and the cantilever beam will deform under the action of this inertial force, causing the piezoresistive film fixed to the cantilever beam to also deform, and the resistance value of the piezoresistive film will change, and the voltage value at both ends will change. Through experiments, a series of relationships between voltage and inertial silicon force are obtained. The inertial force is caused by external acceleration, so the relationship between voltage and acceleration can be obtained to complete the measurement of acceleration. Figure 3 It is based on the principle of silicon piezoresistive micro-acceleration sensor.

[0133] The principle, structure and manufacturing process of piezoresistive micro-accelerometer are simple, and the interface and internal circuit are easy to implement. However, it is very sensitive to temperature changes, which affects the measurement accuracy. The sensitivity is relatively low, which is not convenient for measuring small acceleration changes.

[0134] (2) Capacitive micro-accelerometer

[0135] From the knowledge of electrical engineering, we know that the change of capacitance is related to the distance between the two plates. Therefore, the change of the distance between the two plates can be reflected by the change of capacitance. The change of displacement is deduced from the change of capacitance, and then the acceleration value can be obtained by differential operation.

[0136] The mass block is fixed to the substrate, one plate of the capacitive micro-accelerometer capacitor is fixed to the moving mass block, and the other plate is fixed to the fixed substrate. When there is acceleration, the mass block is displaced, causing the upper and lower capacitors to change, and the acceleration is obtained according to the difference in capacitance change. The capacitive micro-accelerometer has high sensitivity and measurement accuracy, good stability, small temperature drift, and low power consumption. However, the output circuit is complex and easily affected by parasitic capacitance and electromagnetic interference.

[0137] (3) Piezoelectric micro-accelerometer

[0138] When a piezoelectric body is subjected to an external mechanical force, it is polarized, and bound charges of opposite signs appear on the surfaces of both ends of the piezoelectric body. The charge density is proportional to the external mechanical force. This phenomenon is called the direct piezoelectric effect. When a piezoelectric body is subjected to an external electric field, it is deformed, and the deformation is proportional to the strength of the external electric field. This phenomenon is called the inverse piezoelectric effect. Crystals with piezoelectric effect are called piezoelectric crystals. Commonly used piezoelectric crystals include quartz and piezoelectric ceramics. The working principle of the piezoelectric micro-accelerometer is: a layer of piezoelectric material film is covered on the elastic beam. When an external acceleration acts on the mass block, the elastic beam will deform under the action of inertia, and a voltage will be generated between the upper and lower electrodes of the device structure. The mathematical model conversion formula is determined by measuring the change in voltage, thereby obtaining the change in acceleration.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the above embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for applying Star Flash to a MEMS accelerometer integrated with an AI chip, characterized in that: include: S1, using a short-distance communication protocol to locate the source of wireless signals in a sensor environment where MEMS accelerometers are densely deployed; using a long-distance communication protocol for radio wave transmission to send and receive data, mapping out a variety of radio wave signal attenuation models, wherein the radio wave signal attenuation model is a model of the impact of different weather conditions on the strength of terahertz signals; S2, deploying an AI chip on the sensor, and using the AI ​​chip to build a pulse neuron model, the pulse neuron model is used to analyze and predict the changes of the input and output voltage data received by the sensor, and sort the relationship between the multi-pulse input synapses and the excitation thresholds of multiple input commands requiring acceleration and deceleration at the same time by probability value, and output the command according to the high and low ranking of the predicted values ​​to determine the trend relationship between the output voltage and the acceleration; S21, constructing a pulse neuron model using the AI ​​chip, wherein the neuron network of the pulse neuron model includes a pulse input point, a synaptic point, an excitation function, an output point, and a reset time; S22, using the trend between the Markov chain output voltage and acceleration to predict the probability of the next acceleration trend of the inertial sensor device, sorting the relationship between the multi-pulse input synapses and the excitation thresholds of multiple simultaneous input commands requiring acceleration and deceleration by probability value, and giving priority to outputting commands with high prediction values.

2. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 1, characterized in that: In S1, the short-range communication protocol adopts the IEEE802.15.3a channel model.

3. The method of applying Star Flash to a MEMS accelerometer integrated AI chip according to claim 2, characterized in that: The IEEE802.15.3a channel model is expressed as: Where X represents the log-normal shadow attenuation; a k，l represents the attenuation coefficient of the kth path in the lth cluster; L represents the number of clusters; l represents the index of L; K represents the number of multipaths in each cluster; k represents the index of K; T l represents the arrival time of the lth cluster; τ k,l represents the delay of the kth path in the lth cluster relative to the first path; t represents time, and δ() represents the time function of channel attenuation; Among them, a k，l Defined as; a k,l =p k,l b k,l Among them, p k，l represents a discrete random variable with equal probability of taking +1 and -1; β k，l It indicates that the channel coefficient of the k-th path in the l-th cluster follows the log-normal distribution.

4. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 1, characterized in that: In S1, the radio wave signal attenuation model includes: an attenuation value model of rain communication signals, an attenuation value model of cloud and fog communication signals, an attenuation value model of snowfall communication signals and an attenuation value model of sandstorm climate characteristics.

5. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 4, characterized in that: The attenuation value model of the rain communication signal is used to characterize the relationship between the wireless signal attenuation rate and rainfall, wherein the attenuation value γ of the rain communication signal is R The model is: γ R =KR A Among them, R is the rainfall; K and A are parameters in two different polarization conditions: horizontal and vertical.

6. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 1, characterized in that: The attenuation value model of the cloud-fog communication signal is used to characterize the relationship between the wireless signal attenuation rate and visibility, where the attenuation value γ of the cloud-fog communication signal is c&f The model is: Where f is the operating frequency, V m For visibility.

7. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 1, characterized in that: The attenuation value model of the snowfall communication signal is used to characterize the relationship between the attenuation rate of the wireless signal and the change of the snowfall intensity, wherein the attenuation value γ of the snowfall communication signal s The model is: c s =7.47×10 -5 f×I(1+5.77×10 -5 f 3 I 0.6 ) Among them, f is the operating frequency and I is the snowfall intensity.

8. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 4, characterized in that: The dust climate characteristic attenuation value model is used to characterize the relationship between the wireless signal attenuation rate and dust, wherein the dust climate characteristic attenuation value L R The model is: Where ε' and ε" are the real and imaginary parts of the complex dielectric constant of wet dust, respectively; N0 is the volume density of naturally and artificially formed dust samples; m is the mean of lnD; σ is the standard deviation of lnD; f is the frequency; lnD is the logarithm of the dust particle size D.

9. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 1, characterized in that: In S22, the probability of the trend between the historical output voltage and acceleration, the probability of the current trend between the output voltage and acceleration changing from acceleration to deceleration, and the probability of the current trend between the output voltage and acceleration changing from deceleration to acceleration are predicted to predict the probability that the next trend between the output voltage and acceleration will change from acceleration to deceleration, and the probability that the next trend between the output voltage and acceleration will change from deceleration to acceleration.

10. The method of applying Star Flash to a MEMS accelerometer integrated with an AI chip according to claim 1, characterized in that: The MEMS accelerometers include the following types: piezoresistive micro-accelerometers, capacitive micro-accelerometers and piezoelectric micro-accelerometers.

Images