Method for predicting actuation amount data of object part based on shape memory alloy

Abstract

The invention provides a shape memory alloy (SMA)-based method for predicting actuation amount data of an object part. The method comprises the following steps: acquiring electrical signal data of SMA on line; based on a predetermined hysteresis model and the electrical signal data obtained online, the actuation amount data of the object component is predicted, and the step of determining the hysteresis model comprises the steps that multiple sets of sample data corresponding to multiple sampling moments are obtained, and each set of sample data comprises the electrical signal data of the SMA and the actuation amount data of the corresponding object component; based on the input polynomial function, hysteresis model fitting is carried out on the electrical signal data V (t) and the actuation amount data Y (t), and a hysteresis model is determined. According to the method for predicting the actuating amount data of the object part, the accuracy of the hysteresis model is improved, so that the actuating amount data can be predicted more accurately, and the optical anti-shake effect is improved.

Description

Technical Field

[0001] The present application relates to the field of shape memory alloys, and in particular to a method for predicting actuation data of an object component based on shape memory alloys. Background Art

[0002] Shape memory alloy (SMA) is a special metal material that can undergo plastic deformation within a certain temperature range and then restore its original macroscopic shape within another temperature range.

[0003] Since SMA has functional characteristics such as approximate momentum, large driving force output and preset shape memory recovery, and has the advantages of low power consumption drive, no interference noise, light weight and no pollution, smart devices made of SMA have been widely used in aerospace, mechanical electronics, biomedicine, bridge construction, automotive industry and daily life.

[0004] However, the actuation ability of such devices depends on the amount of metal deformation caused by the mutual transformation of low-temperature martensite and high-temperature austenite of the material itself when the metal temperature changes, and the physical properties of heating and cooling are relatively complex, and there are hysteresis characteristics of heating and cooling. The existing hysteresis model is difficult to accurately reflect the actuation amount of such actuators by measuring the electrical signal of SMA. Summary of the invention

[0005] Some embodiments of the present application provide a method for determining a hysteresis model based on SMA and a method for predicting actuation data of an object component. The present application is introduced from multiple aspects below, and the embodiments and beneficial effects of the following aspects can be referenced to each other.

[0006] In a first aspect, an embodiment of the present application provides a method for predicting actuation data of an object component based on a shape memory alloy (SMA), comprising:

[0007] Get the electrical signal data of SMA online;

[0008] Based on a predetermined hysteresis model and the electrical signal data obtained online, the actuation amount data of the target component is predicted, wherein the step of determining the hysteresis model includes:

[0009] Acquire multiple sets of sample data corresponding to multiple sampling moments, each set of sample data including electrical signal data of the SMA and actuation amount data of the corresponding object component;

[0010] Based on the input polynomial function, the hysteresis model is fitted to the electrical signal data and the actuation amount data to determine the hysteresis model.

[0011] The method for predicting actuation quantity data provided in the present application uses the electrical signal data obtained online as the input data of the hysteresis model and the output data of the hysteresis model as the predicted value of the actuation quantity data, thereby realizing accurate prediction of the actuation quantity data based on a hysteresis model with higher accuracy.

[0012] In some embodiments, the step of predicting the actuation amount data of the target component based on the predetermined hysteresis model and the electrical signal data obtained online includes:

[0013] Performing normalization processing on the electrical signal data acquired online to determine the electrical signal data after the normalization processing;

[0014] predicting normalized actuation amount data of the target component based on a predetermined hysteresis model and normalized electrical signal data;

[0015] The normalized actuation amount data is denormalized to determine the actuation amount data.

[0016] In some embodiments, the step of fitting the hysteresis model to the electrical signal data and the actuation amount data based on the input polynomial function to determine the hysteresis model includes:

[0017] Performing normalization processing on multiple groups of sample data to determine normalized electrical signal data and actuation quantity data;

[0018] Based on the input polynomial function, the hysteresis operator, the hysteresis operator coefficient, and the hysteresis operator threshold, the hysteresis model is fitted to the normalized electrical signal data and the actuation amount data to determine the hysteresis model.

[0019] In some embodiments, based on the input polynomial function, the hysteresis operator, the hysteresis operator coefficient, and the hysteresis operator threshold, the normalized electrical signal data and the actuation amount data are fitted with a hysteresis model to determine the hysteresis model, comprising:

[0020] Based on the input polynomial function D(v(t)), the hysteresis operator f(t), the hysteresis operator coefficient p, and the hysteresis operator threshold r j , the normalized electrical signal data v(t) and the actuation data y(t) are fitted with a hysteresis model to establish a hysteresis model, wherein the hysteresis operator f(t) at the current moment is configured based on the electrical signal data v(t) at the current moment, the input term coefficient γ L , γ R , hysteresis operator threshold r j , and the hysteresis operator f(t - )

[0021] Based on the particle swarm optimization algorithm, the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ are determined. L , γ R , to determine the hysteresis model.

[0022] In some embodiments, based on the particle swarm optimization algorithm, the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ are determined. L , γ R The steps include:

[0023] The coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ L , γ R as the parameter vector to be identified;

[0024] Randomly initialize a preset number of parameter vectors to be identified;

[0025] Based on the loss function, determining the individual optimal solution and the group optimal solution of the parameter vectors to be identified in a preset number respectively;

[0026] Iteratively updating the parameter vector to be identified based on the individual optimal solution and the group optimal solution of the parameter vector to be identified;

[0027] Based on the loss function, determine the individual optimal solution and the group optimal solution of the parameter vector to be identified after iterative update until the iteration ends;

[0028] Based on the group optimal solution among all the iteration results, determine the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ L , γ R .

[0029] In some embodiments, the loss function includes a mean square error loss function, a cross entropy loss function, and / or an L2 loss function.

[0030] In some embodiments, the electrical signal data v(t) at the current moment is greater than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is given by the input coefficient γ L , hysteresis operator threshold r j , the electrical signal data v(t) at the current moment and the hysteresis operator f(t - ) is determined; the electrical signal data v(t) at the current moment is equal to the electrical signal data v(t) at the previous moment - ), the hysteresis operator f(t) at the current moment is the hysteresis operator f(t) at the previous moment. -) is determined; the electrical signal data v(t) at the current moment is less than the electrical signal data v(t) at the previous moment - ), the hysteresis operator f(t) at the current moment is given by the input coefficient γ R , hysteresis operator threshold r j , the electrical signal data v(t) at the current moment and the hysteresis operator f(t - ) is determined.

[0031] In some embodiments, the electrical signal data v(t) at the current moment is greater than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is f(t) = max(γ L (v(t))-r j ,f(t - )) is determined, where max(γ L (v(t))-r j ,f(t - )) indicates that the hysteresis operator f(t) at the current moment is γ L (v(t))-r j 、f(t - ), γ L represents the input coefficient, r j represents the hysteresis operator threshold, f(t - ) represents the hysteresis operator at the previous moment; the electrical signal data v(t) at the current moment is equal to the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is f(t)=f(t - ) is determined, where f(t - ) represents the hysteresis operator at the previous moment; the electrical signal data v(t) at the current moment is less than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is f(t) = min(γ R (v(t))+r j ,f(t - )) is determined, where min(γ R (v(t))+r j ,f(t - )) indicates that the hysteresis operator f(t) at the current moment is γ R (v(t))+r j 、f(t - ), γ R represents the input coefficient, r j represents the hysteresis operator threshold, f(t -) represents the hysteresis operator at the previous moment.

[0032] In some embodiments, the electrical signal data of the SMA includes resistance signal data, current signal data, and / or power signal data of the SMA.

[0033] In some embodiments, the SMA includes a first SMA and a second SMA, one end of the first SMA and the second SMA are respectively connected to the object component, the other ends of the first SMA and the second SMA are respectively fixed, and the signal data of the difference in resistance between the first SMA and the second SMA is used as the electrical signal data.

[0034] In a second aspect, the present application also provides a chip, the chip comprising a processor and a data interface, the processor reads instructions stored in a memory through the data interface to execute any of the above methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A partial structural schematic diagram of an SMA-based actuator provided according to some embodiments of the present application is shown;

[0036] Figure 2 A schematic diagram showing a device for predicting actuation amount data according to some embodiments of the present application;

[0037] Figure 3a A flow chart showing a method for predicting actuation amount data provided according to some embodiments of the present application;

[0038] Figure 3b A schematic diagram showing a normalization process of electrical signal data R determined by formula (1) provided in some embodiments of the present application is shown;

[0039] Figure 3c A schematic diagram showing a process of normalizing the actuation amount data X determined by formula (2) provided in some embodiments of the present application;

[0040] Figure 4 A flowchart showing a method for determining a hysteresis model according to some embodiments of the present application;

[0041] Figure 5 A flowchart showing a method for determining a hysteresis model according to some other embodiments of the present application;

[0042] Figure 6 Show Figure 5 A flowchart of an implementation of step S51;

[0043] Figure 7 A flowchart showing a method for determining parameters based on a particle swarm optimization algorithm provided according to some embodiments of the present application;

[0044] Figure 8 A schematic diagram showing a method for predicting actuation amount data according to some embodiments of the present application;

[0045] Fig. 9 A block diagram of a SoC (System on Chip) provided according to some embodiments of the present application is shown. DETAILED DESCRIPTION

[0046] The specific implementation of the present application will be described in detail below with reference to the accompanying drawings.

[0047] For ease of understanding, the local mechanical structure of the SMA-based actuator is first exemplarily introduced. Figure 1 FIG. 1 is a schematic diagram showing a partial structure of an SMA-based actuator according to some embodiments of the present application. Figure 1 As shown, the actuator comprises: two SMA metal wires (such as Figure 1 The two SMA wires are connected to the mass blocks at one end and fixed at the other end (for example, fixed to the base or support of the mass blocks). When the SMA wires are heated, they can generate a contraction force, thereby driving the mass blocks to move (for example, along the Figure 1 Therefore, by heating SMA1 and SMA2 alternately, the mass block can be heated in this degree of freedom (such as Figure 1 Reciprocating motion in the horizontal direction).

[0048] It should be noted that the "opposite arrangement" of SMA wires refers to the opposite arrangement in the mechanical sense, rather than just the opposite arrangement in the structural sense. In other words, the arrangement of the SMA wires can drive the mass block to move in a single degree of freedom after being heated, and this application does not impose any specific restrictions on this.

[0049] Although only a single-degree-of-freedom actuator structure is shown above, those skilled in the art will appreciate that a multi-degree-of-freedom actuator structure may be a mechanical structure integrating multiple single-degree-of-freedom structures described above, which will not be described in detail here.

[0050] The following is a further introduction using the application of SMA-based actuators in the field of optical image stabilization (OIS) of electronic devices as an example.

[0051] The electronic device may specifically be a mobile phone, tablet computer, wearable device, laptop computer, ultra-mobile personal computer (UMPC), handheld computer, netbook, personal digital assistant (PDA), virtual reality device, or other electronic device with a shooting function, and this application does not impose any specific restrictions on this.

[0052] The mass block (or "object component") is equivalent to the camera in an electronic device with a shooting function. When using the shooting function of an electronic device (such as a mobile phone), the camera often shakes due to operator hand shaking or external bumps, resulting in blurred and unclear images. Therefore, it is necessary to further adjust the position of the camera to offset the camera shaking caused by hand shaking or external bumps, so as to achieve the purpose of optical image stabilization.

[0053] Combination Figure 1 As shown, when the operator's hand shakes, the mass block (camera), SMA1 and SMA2 will move. Since the displacements of the mass block and the SMA wire are different, a relative displacement will occur between the SMA wire and the mass block, causing the contraction length of the SMA wire to change, that is, the electrical signal of the SMA wire to change. Therefore, the electrical signal data of the SMA wire can be obtained to determine the position change signal (that is, the actuation data) of the mass block (camera), so that the temperature of the SMA wire can be adjusted (for example, specifically by adjusting the current of the SMA wire) to cause the SMA wire to deform to generate a contraction force, thereby driving the mass block (camera) to have a preset displacement to offset its position change caused by shaking and improve the clarity of the image.

[0054] Therefore, the accuracy of determining the actuation data of the mass block (or “object component”) based on the electrical signal data of the SMA wire largely determines the effect of the optical image stabilization.

[0055] The present application proposes a method and device for predicting actuation quantity data, which is based on a more accurate hysteresis model, thereby being able to more accurately predict the actuation quantity data.

[0056] Figure 2 FIG. 2 is a schematic diagram showing a device for predicting actuation amount data according to some embodiments of the present application. Figure 2As shown, the predicted actuation amount data device 1 includes: an online electrical signal data acquisition unit 10 and a predicted actuation amount data acquisition unit 30. The online electrical signal data acquisition unit 10 is configured to acquire the electrical signal data of the SMA online. The electrical signal data may include resistance signal data, current signal data and / or power signal data of the SMA. For example, the electrical signal data is resistance signal data, which may be Figure 1 , which is signal data of the difference in resistance between SMA1 (first SMA) and SMA2 (second SMA) shown in FIG.

[0057] The predicted actuation amount data unit 30 is configured to predict the target component (such as Figure 1 The actuation data of the mass block shown in FIG. How to determine the hysteresis model will be described in detail below.

[0058] In other words, the device 1 for predicting actuation amount data provided in the present application uses the electrical signal data acquired online as the input data of the hysteresis model and the output data of the hysteresis model as the predicted value of the actuation amount data to achieve accurate prediction of the actuation amount data.

[0059] In some embodiments, the device for predicting actuation amount data further includes: a data normalization processing unit 20. The data normalization processing unit 20 is configured to perform normalization (or "standardization") processing on the electrical signal data acquired by the online electrical signal data acquisition unit 10.

[0060] In some embodiments, the device for predicting actuation amount data further includes: a data denormalization processing unit 40. The data denormalization processing unit 40 is configured to perform a denormalization process on the normalized actuation amount data predicted by the predicting actuation amount data unit 30.

[0061] Figure 3a A flow chart showing a method for predicting actuation amount data according to some embodiments of the present application is shown. Figure 3a As shown, the method comprises the following steps:

[0062] Step S31, acquiring electrical signal data of the SMA online.

[0063] It is understandable that after the optical image stabilization function is turned on in an electronic device (such as a mobile phone), when the gyroscope in the electronic device detects that the electronic device is shaking, the analog-to-digital converter is triggered to detect the current electrical signal data of the SMA as the electrical signal data of the SMA obtained online. The electrical signal data may include resistance signal data, current signal data and / or power signal data of the SMA. Exemplarily, the electrical signal data is resistance signal data.

[0064] Step S32, normalizing the electrical signal data of the SMA obtained online to determine the electrical signal data after the normalization process.

[0065] For example, the electrical signal data of the SMA obtained online is R, which can be normalized by the following formula:

[0066]

[0067] In formula (1), R std represents the electrical signal data after normalization (standardization); R represents the electrical signal data before normalization (standardization); R max , R min They respectively represent the maximum and minimum values ​​of the electrical signal data before normalization. Figure 3b FIG. 2 shows the normalization process of the electrical signal data R determined by formula (1) provided in some embodiments of the present application. Figure 3b As shown, the normalization process performed by formula (1) can make the normalized electrical signal data R std The range of is [-0.5, 0.5]. In some other embodiments, the electrical signal data can be normalized by other methods, for example, std The range is [0,1], [-1,0], and is not specifically limited here, as long as the data span is 1.

[0068] In step S33, based on a predetermined hysteresis model and the normalized electrical signal data, the normalized actuation amount data of the target component is predicted.

[0069] For example, if the electrical signal data of the SMA obtained online is normalized by the above formula (1), then based on the predetermined hysteresis model and the normalized electrical signal data R std The predicted actuation amount data X after normalization is determined std , which satisfies similar formula (1), as follows:

[0070]

[0071] In formula (2), X represents the actuation data after denormalization; X std represents the actuation data after normalization; X max , X min They respectively represent the maximum value and the minimum value of the actuation amount data before normalization processing.

[0072] Figure 3c The normalization process of the actuation amount data X determined by formula (2) provided in some embodiments of the present application is shown. Figure 3c As shown, the actuation data X before denormalization std The range is [-0.5, 0.5].

[0073] Among them, the predetermined hysteresis model improves the accuracy of the hysteresis model compared with the existing hysteresis model by adding an input polynomial function and / or adding input term coefficients to the hysteresis operator, thereby being able to more accurately predict the actuation data and better achieve the optical image stabilization effect.

[0074] Step S34, denormalizing the normalized actuation amount data to determine the actuation amount data of the target component.

[0075] By formula (2), we can get:

[0076] X=(X std +0.5)(X max -X min )+X min Formula (3)

[0077] In formula (3), X represents the actuation data after denormalization; X std represents the actuation data after normalization (de-standardization); X max , X min They respectively represent the maximum value and the minimum value of the actuation amount data before normalization processing.

[0078] It should be noted that for the parts not described in detail in the steps, please refer to the above Figure 2 The description in will not be repeated here.

[0079] In other words, the device and method for predicting actuation quantity data provided in the present application use the electrical signal data acquired online as the input data of the hysteresis model and the output data of the hysteresis model as the predicted value of the actuation quantity data to achieve accurate prediction of the actuation quantity data.

[0080] The following specifically introduces the method for determining the hysteresis model mentioned above.

[0081] Figure 4 FIG. 1 is a flow chart showing a method for determining a hysteresis model according to some embodiments of the present application. Figure 4 As shown, the method for determining the hysteresis model includes the following steps:

[0082] Step S41 , obtaining a plurality of groups of sample data corresponding to a plurality of sampling times, each group of sample data including electrical signal data V(t) of the SMA and actuation amount data Y(t) of the corresponding object component.

[0083] It should be noted that the sample data may be sample data corresponding to the sampling time in any time period. For example, each sample data may be sample data corresponding to the number of sampling times, that is, the electrical signal data of the SMA collected by the first sampling is recorded as V(1), and the actuation data of the object component is recorded as Y(1); the electrical signal data of the SMA collected by the second sampling is recorded as V(2), and the actuation data of the object component is recorded as Y(2), ..., the electrical signal data of the SMA collected by the lth sampling is recorded as V(l), and the actuation data of the object component is recorded as Y(l).

[0084] Step S42: Based on the input polynomial function, a hysteresis model is fitted to the electrical signal data V(t) and the actuation amount data Y(t) to determine the hysteresis model.

[0085] It should be noted that the input polynomial function is not specifically limited here. For example, they can all be even powers, for example, f(x)=a+bx 2 +cx 4 +dx 6 ; They can also be odd powers, for example, f(x) = ax + bx 3 +cx 5 ; It can also be a combination of the two, for example, f(x) = ax + bx 2 +cx 4 +dx 7 .

[0086] According to the method for determining a hysteresis model based on SMA provided in the present application, the accuracy of the hysteresis model is improved by adding an input polynomial function, so as to achieve a more accurate prediction of the actuation quantity data.

[0087] In other embodiments, to facilitate the fitting of the hysteresis model, multiple groups of sample data may be normalized, and then the hysteresis model may be fitted based on the normalized electrical signal data v(t) and actuation data y(t).

[0088] Figure 5 FIG. 1 is a flow chart showing a method for determining a hysteresis model according to some other embodiments of the present application. Figure 5As shown, the method for determining the hysteresis model includes the following steps:

[0089] Step S51 , obtaining a plurality of groups of sample data corresponding to a plurality of sampling times, each group of sample data including electrical signal data V(t) of the SMA and actuation amount data Y(t) of the corresponding object component.

[0090] Step S52, normalizing the plurality of groups of sample data to determine the normalized electrical signal data v(t) and actuation amount data y(t).

[0091] By normalizing the electrical signal data V(t) and the actuation data Y(t), the data ranges of the two are unified within a certain data range, for example Figure 3b and Figure 3c The [-0.5, 0.5] shown may also be [0, 1] or [-1, 0], and is not specifically limited here, as long as the data span is 1.

[0092] Taking the data range [-0.5, 0.5] as an example, the normalization processing of the electrical signal data V(t) and the actuation data Y(t) can be achieved by the following formula:

[0093]

[0094] In formula (4), v(t) represents the electrical signal data after normalization (standardization); y(t) represents the actuation data after normalization (standardization); V(t) represents the electrical signal data before normalization (standardization); Y(t) represents the actuation data before normalization (standardization); V max 、V min Respectively represent the maximum and minimum values ​​of the electrical signal data before normalization (standardization); Y max , Y min They respectively represent the maximum value and the minimum value of the actuation amount data before normalization processing.

[0095] Step S53, based on the input polynomial function, the hysteresis operator, the hysteresis operator coefficient, and the hysteresis operator threshold, a hysteresis model is fitted to the normalized electrical signal data v(t) and the actuation amount data y(t) to determine the hysteresis model.

[0096] The hysteresis operator, the hysteresis operator coefficient, and the hysteresis operator threshold will be introduced in detail below.

[0097] In some embodiments, Figure 6 As shown, step S53 may specifically include the following steps:

[0098] Step S531, based on the input polynomial function, the hysteresis operator, the hysteresis operator coefficient, and the hysteresis operator threshold, a hysteresis model is fitted to the normalized electrical signal data v(t) and the actuation data y(t) to establish a hysteresis model;

[0099] The hysteresis model can be established by the following formula:

[0100]

[0101] In formula (5), v(t) represents the normalized electrical signal data (input); y(t) represents the normalized actuation data (output); D(v(t)) represents the input polynomial function, and the structure of the input polynomial function is generally determined based on experience; N represents the hysteresis model order, generally N ≥ 20; p represents the hysteresis operator coefficient; r j Represents the hysteresis operator threshold, the value range is 0≤r1 <r2<…r N <∞, generally based on experience, its maximum value is equal to the maximum value of the electrical signal data after normalization; f(t), represents the hysteresis operator, which can be determined by the following formula:

[0102]

[0103] In formula (6), γ L , γ R represents the input coefficient; v(t) represents the electrical signal data at the current moment after normalization; r j represents the hysteresis operator threshold, f(t - ) represents the hysteresis operator at the previous moment.

[0104] In other words, the electrical signal data v(t) at the current moment is greater than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is given by Determine, which means that the hysteresis operator f(t) at the current moment is γ L (v(t))-r j 、f(t - ) is the maximum value; the electrical signal data v(t) at the current moment is equal to the electrical signal data v(t) at the previous moment - ), the hysteresis operator f(t) at the current moment is given by Determine; the electrical signal data v(t) at the current moment is less than the electrical signal data v(t at the previous moment - ), the hysteresis operator f(t) at the current moment is given by Determine, which means that the hysteresis operator f(t) at the current moment is γ R (v(t))+r j 、f(t - ) is the minimum value among .

[0105] From the above formulas (5) and (6), we can see that the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ L , γ R These are the undetermined parameters (or “parameters to be identified”). The following steps will specifically introduce how to determine these parameters.

[0106] Step S532: Determine the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ based on the particle swarm optimization algorithm. L , γ R , to determine the hysteresis model.

[0107] According to the method for determining a hysteresis model based on SMA provided in the present application, a particle swarm optimization algorithm (PSO) is used to determine the parameters to be identified in the hysteresis model, thereby further improving the accuracy of the hysteresis model.

[0108] Figure 7 A flow chart showing a method for determining parameters based on a particle swarm optimization algorithm according to some embodiments of the present application is shown. Figure 7 , specifically introduces how to determine parameters based on the particle swarm optimization algorithm. Determining parameters based on the particle swarm optimization algorithm includes the following steps:

[0109] Step S5321, the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ L , γ R as the parameter vector to be identified.

[0110] Among them, the parameter vector to be identified is recorded as, x id =(x i1 ,x i2 ,…,x id ), i.e., a 1×d matrix, where d represents the number of parameters to be identified, i.e., the coefficients of the polynomial function D(v(t)), the hysteresis operator coefficients p, and the input term coefficients γ L , γ R The total number of .

[0111] Step S5322: randomly initialize a preset number of parameter vectors to be identified. The preset number is generally an empirical value, such as 30 or 40.

[0112] Step S5323, based on the loss function, determine the individual optimal solution and the group optimal solution of the parameter vectors to be identified in a preset number.

[0113] Among them, the loss function may include a mean square error loss function, a cross entropy loss function and / or an L2 loss function, etc.

[0114] In this embodiment, the loss function is a mean square error loss function, which can achieve similar effects while having a simpler algorithm than other functions. The specific formula is as follows:

[0115]

[0116] In formula (7), k represents the data length (iteration number), k = 1, 2, 3, ...; loss[k] represents the error scalar at the kth iteration; y[i] represents the true value, that is, the actuation data after normalization; y predict [i] represents the normalized actuation data predicted by the above hysteresis model.

[0117] In this embodiment, k is equal to 1, that is, the kth iteration is the lth sampling. In other embodiments, the two may not be equal, as long as the total number of iterations is greater than the number of samplings.

[0118] Step S5324, iteratively update the parameter vector to be identified based on the individual optimal solution and the group optimal solution of the parameter vector to be identified.

[0119] The iterative update of the parameter vector to be identified can be specifically performed through the following formula:

[0120]

[0121] In formula (8), x id represents the parameter vector to be identified; v id represents the optimization direction of the parameter vector to be identified; k represents the kth iteration; ω represents the inertia weight, which is generally an empirical value; c1 and c2 represent the learning factors, which are generally empirical values; r rand1 、r rand2 Represents a random number in the interval [0,1] to increase the randomness of the search; represents the individual optimal solution of the parameter vector to be identified at the kth iteration; It represents the group optimal solution of the parameter vector to be identified in the kth iteration, that is, the optimal solution among all individual optimal solutions.

[0122] Step S5325, based on the loss function, determine the individual optimal solution and the group optimal solution of the parameter vector to be identified after iterative update, until the iteration ends.

[0123] In other words, step S5324 and step S5325 are repeated until the iteration ends.

[0124] Step S5326: Based on the group optimal solution in all iteration results, determine the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ L , γ R , to determine the hysteresis model.

[0125] The hysteresis model determined by the method for determining the hysteresis model provided by the present application is increased by adding the input polynomial function D(v(t)) and / or the input term coefficient γ L , γ R , improving the accuracy of the hysteresis model, thereby improving the accuracy of the actuation data prediction.

[0126] Figure 8 A schematic diagram of a method for predicting actuation amount data according to some embodiments of the present application is shown.

[0127] First, please refer to Figure 8 The upper dashed box 701 is used to introduce how to determine the hysteresis model.

[0128] (1) acquiring multiple sets of sample data corresponding to multiple sampling times, each set of sample data including SMA electrical signal data V(t) and corresponding mass block displacement data (or “actuation amount data of the object component”) Y(t);

[0129] (2) performing data standardization (or “normalization”) on the above-mentioned SMA electrical signal data V(t) and mass block displacement data Y(t);

[0130] (3) fitting the hysteresis model of the standardized electrical signal data v(t) and the mass block displacement data y(t) to establish a hysteresis model;

[0131] (4) Parameter identification is performed on the parameters to be identified in the hysteresis model. The parameter identification can be determined by the particle swarm optimization algorithm in the above embodiment, that is, the parameter weights are determined, and finally the hysteresis model is determined.

[0132] Secondly, refer to Figure 8 The remaining part will introduce how to predict the displacement data of the mass block (the actuation amount data of the object component).

[0133] (1) Obtain SMA electrical signal data online;

[0134] (2) Standardize the SMA electrical signal data obtained online;

[0135] (3) predicting the normalized mass displacement data (actuation amount data of the target component) based on the hysteresis model determined by the dashed box 701 and the normalized electrical signal data;

[0136] (4) De-standardizing (de-normalizing) the normalized mass displacement data (actuation amount data of the target component) to determine the mass displacement data (actuation amount data of the target component).

[0137] It should be noted that for steps not specifically described, please refer to the above embodiments and will not be repeated here.

[0138] Fig. 9 A block diagram of a SoC (System on Chip) provided according to some embodiments of the present application is shown. Fig. 9 In the FIG, similar components have the same reference numerals. In addition, the dashed boxes are optional features of more advanced SoCs. Fig. 9 In the embodiment, SoC 1500 includes: an interconnect unit 1550, which is coupled to an application processor 1515; a system agent unit 1570; a bus controller unit 1580; an integrated memory controller unit 1540; a group or one or more coprocessors 1520, which may include integrated graphics logic, image processors, audio processors, and video processors; a static random access memory (SRAM) unit 1530; and a direct memory access (DMA) unit 1560. In one embodiment, the coprocessor 1520 includes a special-purpose processor, such as, for example, a network or communication processor, a compression engine, a GPGPU, a high-throughput MIC processor, or an embedded processor.

[0139] The various embodiments of the mechanism disclosed in the present application can be implemented in hardware, software, firmware or a combination of these implementation methods. The embodiments of the present application can be implemented as a computer program or program code executed on a programmable system, which includes at least one processor, a storage system (including volatile and non-volatile memory and / or storage elements), at least one input device and at least one output device.

[0140] Program code can be applied to input instructions to perform the functions described in this application and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, a processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application specific integrated circuit (ASIC), or a microprocessor.

[0141] Program code can be implemented with high-level programming language or object-oriented programming language to communicate with the processing system. When necessary, program code can also be implemented with assembly language or machine language. In fact, the mechanism described in this application is not limited to the scope of any specific programming language. In either case, the language can be a compiled language or an interpreted language.

[0142] In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried or stored on one or more temporary or non-temporary machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, instructions may be distributed over a network or through other computer-readable media. Therefore, machine-readable media may include any mechanism for storing or transmitting information in a machine (e.g., computer) readable form, including, but not limited to, floppy disks, optical disks, optical disks, read-only memories (CD-ROMs), magneto-optical disks, read-only memories (ROMs), random access memories (RAMs), erasable programmable read-only memories (EPROMs), electrically erasable programmable read-only memories (EEPROMs), magnetic or optical cards, flash memory, or a tangible machine-readable memory for transmitting information (e.g., carrier waves, infrared signals, digital signals, etc.) using the Internet in electrical, optical, acoustic, or other forms of propagation signals. Therefore, machine-readable media include any type of machine-readable media suitable for storing or transmitting electronic instructions or information in a machine (e.g., computer) readable form.

[0143] In the accompanying drawings, some structural or method features may be shown in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order may not be required. Instead, in some embodiments, these features may be arranged in a manner and / or order different from that shown in the illustrative drawings. In addition, the inclusion of structural or method features in a particular figure does not mean that such features are required in all embodiments, and in some embodiments, these features may not be included or may be combined with other features.

[0144] It should be noted that the units / modules mentioned in the various device embodiments of the present application are all logical units / modules. Physically, a logical unit / module can be a physical unit / module, or a part of a physical unit / module, or can be implemented as a combination of multiple physical units / modules. The physical implementation method of these logical units / modules themselves is not the most important. The combination of functions implemented by these logical units / modules is the key to solving the technical problems proposed by the present application. In addition, in order to highlight the innovative part of the present application, the above-mentioned device embodiments of the present application do not introduce units / modules that are not closely related to solving the technical problems proposed by the present application, which does not mean that there are no other units / modules in the above-mentioned device embodiments.

[0145] It should be noted that in the examples and description of this patent, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including one" do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0146] Each method implementation mode of the present application can be implemented in the form of software, magnetic components, firmware, etc.

[0147] Program code can be applied to input instructions to perform the functions described herein and generate output information. The output information can be applied to one or more output devices in a known manner. For purposes of this application, a processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application specific integrated circuit (ASIC), or a microprocessor.

[0148] Program code can be implemented with high-level programming language or object-oriented programming language to communicate with the processing system. When necessary, program code can also be implemented with assembly language or machine language. In fact, the mechanism described herein is not limited to the scope of any specific programming language. In either case, the language can be a compiled language or an interpreted language.

[0149] One or more aspects of at least one embodiment may be implemented by representative instructions stored on a computer-readable storage medium, which represent various logics in a processor, and when the instructions are read by a machine, the machine makes logic for performing the technology herein. These representations, known as "IP cores," may be stored on a tangible computer-readable storage medium and provided to multiple customers or production facilities to be loaded into a manufacturing machine that actually manufactures the logic or processor.

[0150] In some cases, the instruction converter can be used to convert instructions from a source instruction set to a target instruction set. For example, the instruction converter can transform (e.g., using static binary transformation, dynamic binary transformation including dynamic compilation), deform, simulate or otherwise convert instructions into one or more other instructions that will be processed by the IP core. The instruction converter can be implemented with software, hardware, firmware, or a combination thereof. The instruction converter can be on a processor, outside a processor, or partly on a processor and partly outside a processor.

[0151] Although the present application has been illustrated and described with reference to certain preferred embodiments thereof, it will be apparent to those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present application.

Claims

1. A method for predicting actuation data of an object component based on shape memory alloy (SMA), characterized in that: include: Get the electrical signal data of SMA online; Based on a predetermined hysteresis model and the electrical signal data acquired online, predicting the actuation amount data of the object component, wherein the step of determining the hysteresis model comprises: Acquire multiple sets of sample data corresponding to multiple sampling moments, each set of sample data including electrical signal data of the SMA and corresponding actuation amount data of the object component; Based on an input polynomial function, a hysteresis model is fitted to the electrical signal data and the actuation amount data to determine the hysteresis model.

2. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 1, characterized in that: The step of predicting the actuation amount data of the object component based on a predetermined hysteresis model and the electrical signal data acquired online comprises: Normalizing the electrical signal data obtained online to determine the normalized electrical signal data; predicting normalized actuation amount data of the target component based on a predetermined hysteresis model and the normalized electrical signal data; The normalized actuation amount data is denormalized to determine the actuation amount data.

3. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 1, characterized in that: The step of fitting the electrical signal data and the actuation amount data with a hysteresis model based on an input polynomial function to determine the hysteresis model comprises: Normalizing the plurality of groups of sample data to determine normalized electrical signal data and actuation amount data; Based on the input polynomial function, the hysteresis operator, the hysteresis operator coefficient, and the hysteresis operator threshold, a hysteresis model is fitted to the normalized electrical signal data and the actuation amount data to determine the hysteresis model.

4. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 3, characterized in that: The step of performing hysteresis model fitting on the normalized electrical signal data and the actuation amount data based on the input polynomial function, the hysteresis operator, the hysteresis operator coefficient, and the hysteresis operator threshold to determine the hysteresis model comprises: Based on the input polynomial function D(v(t)), the hysteresis operator f(t), the hysteresis operator coefficient p, and the hysteresis operator threshold r j , the normalized electrical signal data v(t) and the actuation data y(t) are fitted with a hysteresis model to establish a hysteresis model, wherein the hysteresis operator f(t) at the current moment is configured to be based on the electrical signal data v(t) at the current moment, the input term coefficient γ L , γ R , the hysteresis operator threshold r j , and the hysteresis operator f(t - ) Based on the particle swarm optimization algorithm, the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ are determined. L , γ R , to determine the hysteresis model.

5. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 4, characterized in that: The particle swarm optimization algorithm is used to determine the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ L , γ R The steps include: The coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ L , γ R as the parameter vector to be identified; Randomly initializing a preset number of parameter vectors to be identified; Based on the loss function, determining the individual optimal solution and the group optimal solution of the preset number of parameter vectors to be identified respectively; Iteratively updating the parameter vector to be identified based on the individual optimal solution and the group optimal solution of the parameter vector to be identified; Based on the loss function, determining the individual optimal solution and the group optimal solution of the parameter vector to be identified after iterative update until the iteration ends; Based on the group optimal solution in all iteration results, the coefficients of the input polynomial function D(v(t)), the hysteresis operator coefficient p, and the input term coefficient γ are determined. L , γ R .

6. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 5, characterized in that: The loss function includes a mean square error loss function, a cross entropy loss function and / or an L2 loss function.

7. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 4, characterized in that: The electrical signal data v(t) at the current moment is greater than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is composed of the input coefficient γ L , hysteresis operator threshold r j , the electrical signal data v(t) at the current moment and the hysteresis operator f(t - ) The electrical signal data v(t) at the current moment is equal to the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is obtained by the hysteresis operator f(t) at the previous moment. - ) The electrical signal data v(t) at the current moment is less than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is composed of the input coefficient γ R , hysteresis operator threshold r j , the electrical signal data v(t) at the current moment and the hysteresis operator f(t - ) is determined.

8. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 7, characterized in that: The electrical signal data v(t) at the current moment is greater than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is f(t)=max(γ L (v(t))-r j ,f(t - )) is determined, where max(γ L (v(t))-r j ,f(t - )) indicates that the hysteresis operator f(t) at the current moment is γ L (v(t))-r j 、f(t - ), γ L represents the input coefficient, r j represents the hysteresis operator threshold, f(t - ) represents the hysteresis operator at the previous moment; The electrical signal data v(t) at the current moment is equal to the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is f(t)=f(t - ) is determined, where f(t - ) represents the hysteresis operator at the previous moment; The electrical signal data v(t) at the current moment is less than the electrical signal data v(t) at the previous moment. - ), the hysteresis operator f(t) at the current moment is f(t)=min(γ R (v(t))+r j ,f(t - )) is determined, where min(γ R (v(t))+r j ,f(t - )) indicates that the hysteresis operator f(t) at the current moment is γ R (v(t))+r j 、f(t - ), γ R represents the input coefficient, r j represents the hysteresis operator threshold, f(t - ) represents the hysteresis operator at the previous moment.

9. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 1, characterized in that: The electrical signal data of the SMA includes resistance signal data, current signal data and / or power signal data of the SMA.

10. The method for predicting actuation amount data of an object component based on shape memory alloy (SMA) according to claim 9, characterized in that: The SMA includes a first SMA and a second SMA, one end of each of the first SMA and the second SMA is connected to the target component, the other ends of each of the first SMA and the second SMA are fixed, and signal data of a difference in resistance between the first SMA and the second SMA is used as the electrical signal data.

11. A chip, characterized in that: The chip includes a processor and a data interface, and the processor reads instructions stored in a memory through the data interface to execute the method according to any one of claims 1 to 10.

Images