Silicon piezoresistive pressure sensor temperature compensation method and system based on IRIME-CNN-DD

Abstract

The invention discloses a silicon piezoresistive pressure sensor temperature compensation method and system based on IRIME-CNN-DD, relates to the technical field of pressure detection, and aims to solve the problem of low temperature compensation accuracy of an existing software compensation method. The method comprises the following steps: firstly, acquiring pressure data at different temperatures as a training data set; carrying out normalization processing on the training data set; the model parameters of the temperature compensation model are optimized through an improved rime algorithm, and the optimal model parameters are obtained; inputting the normalized training data set into a temperature compensation model, and training the temperature compensation model by using the optimal model parameters; and inputting the pressure data to be measured into the trained temperature compensation model for temperature compensation. The improved rime algorithm is used for optimization in the aspect of neural network parameter optimization, the robustness of the model is improved, local features of the temperature compensation data set are extracted through the convolutional neural network, and the sensitivity of the pressure sensor is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pressure detection, and in particular to a temperature compensation method and system for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD. Background Art

[0002] Silicon piezoresistive pressure sensors are widely used in many fields such as environmental monitoring, shipbuilding and marine engineering, and the energy industry due to their low cost, small size, and high sensitivity, and have become a key component to promote technological progress and development in various fields. However, due to the thermal sensitivity of silicon-based semiconductors, silicon piezoresistive pressure sensors are easily affected by factors such as temperature, which can cause errors in the measurement results.

[0003] There are two main methods of temperature compensation currently in common use. The first is the hardware compensation method, which is mainly to directly offset the effects of temperature changes by adding physical components to the system. The second is the software compensation method, which is mainly to indirectly offset the effects of temperature changes by using software algorithms. Compared with hardware compensation strategies, software compensation strategies are easy to adjust and upgrade, and do not increase the physical complexity of the system. Therefore, they have gradually become a research hotspot. Currently, the widely used software compensation methods mainly include neural network algorithms, least squares methods, and interpolation compensation. Since neural networks have stronger multi-dimensional non-mapping capabilities than other software compensation algorithms, this capability is widely used in sensor temperature compensation systems.

[0004] However, existing neural network temperature compensation methods generally have problems such as slow convergence speed and easy to fall into the local minimum of the error function. Especially when dealing with large search spaces, multi-peaks and non-differentiable functions, traditional methods often find it difficult to effectively find the global minimum, and their generalization ability is also limited. These defects have largely restricted the temperature compensation performance of traditional neural network algorithms in silicon pressure sensors. Summary of the invention

[0005] In view of the above problems, the present invention proposes a temperature compensation method and system for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD to solve the problems existing in the existing software compensation method.

[0006] According to one aspect of the present invention, a temperature compensation method for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD is proposed, the method comprising:

[0007] Obtain pressure data at different temperatures as a training data set;

[0008] Normalizing the training data set;

[0009] The improved rime algorithm is used to optimize the model parameters of the temperature compensation model to obtain the optimal model parameters; the model parameters include an initial learning rate and a maximum number of iterations; the temperature compensation model is a temperature compensation model based on a convolutional neural network;

[0010] Inputting the normalized training data set into the temperature compensation model, and training the temperature compensation model using the optimal model parameters to obtain a trained temperature compensation model;

[0011] The pressure data to be measured is input into the trained temperature compensation model for temperature compensation to obtain the corrected pressure data.

[0012] Furthermore, the method of optimizing the model parameters of the temperature compensation model by using the improved rime algorithm to obtain the optimal model parameters includes:

[0013] Initialize the population, fitness function, the global optimal solution of the particle, the historical individual optimal solution of the particle, the maximum number of iterations T and the current number of iterations t;

[0014] According to the fitness function of each particle in the population, the population is divided into a deep mining population and a global search population;

[0015] Iterate the following process:

[0016] Update the positions of particles in the deep mining population according to the deep mining learning strategy position update formula;

[0017] Update the positions of particles in the global search population according to the global exploration learning strategy position update formula;

[0018] A fixed percentage of particles are selected from the deep mining population and the global search population, and a crossover population is generated according to the crossover population position update formula generated by the adaptive Lapp crossover operator;

[0019] Merge the deep mining, global exploration population and cross population to get the merged population;

[0020] Use aggressive greedy updates to merge the global optimal solution and historical individual optimal solutions in the population;

[0021] Adjust the particle ratio between the deep mining population and the global search population according to preset rules;

[0022] When the number of iterations t is greater than or equal to the maximum number of iterations T, the iteration is stopped and the global optimal solution, that is, the optimal model parameters, is output.

[0023] Furthermore, the deep mining learning strategy position update formula is:

[0024]

[0025] In the formula, Represents the position of particles in the deep mining population after update; R best,j represents the best particle; r1 is a random number in the range of (-1,1); t is the current iteration number; T is the maximum iteration number; h represents the adhesion; Ub ij and Lb ij are the upper and lower bounds of the escape space respectively; r2 is a random number in the range of (0,1); trnd(T+t) is a random number that obeys the Student's t-distribution; r4 is a random number in the range of (0,1).

[0026] Furthermore, the global exploration learning strategy position update formula is:

[0027]

[0028] In the formula, Represents the position of particles in the global exploration population after update.

[0029] Furthermore, the crossover population position update formula generated by the adaptive Lapp crossover operator is:

[0030]

[0031] In the formula, is the individual historical optimal solution; c is a random number in the range of [-1,1]; r5 is a random number in the range of [0,1]; r is a random number in the range of [0,1], a is the position parameter, and u is a random number in the range of [0,1] that follows a uniform distribution.

[0032] Furthermore, the fitness function is the mean absolute error.

[0033] Furthermore, the temperature compensation model is a temperature compensation model based on a convolutional neural network and a dendritic network.

[0034] According to another aspect of the present invention, a temperature compensation system for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD is proposed, the system comprising:

[0035] A data acquisition module configured to acquire pressure data at different temperatures as a training data set;

[0036] A preprocessing module, configured to perform normalization processing on the training data set;

[0037] A parameter optimization module, configured to optimize the model parameters of the temperature compensation model using an improved rime algorithm to obtain optimal model parameters; the model parameters include an initial learning rate and a maximum number of iterations; the temperature compensation model is a temperature compensation model based on a convolutional neural network;

[0038] A model training module, configured to input the normalized training data set into the temperature compensation model, and train the temperature compensation model using optimal model parameters to obtain a trained temperature compensation model;

[0039] The temperature compensation module is configured to input the pressure data to be measured into a trained temperature compensation model for temperature compensation to obtain corrected pressure data.

[0040] Furthermore, the parameter optimization module uses an improved rime algorithm to optimize the model parameters of the temperature compensation model, and obtaining the optimal model parameters includes:

[0041] Initialize the population, fitness function, the global optimal solution of the particle, the historical individual optimal solution of the particle, the maximum number of iterations T and the current number of iterations t;

[0042] According to the fitness function of each particle in the population, the population is divided into a deep mining population and a global search population;

[0043] Iterate the following process:

[0044] Update the positions of particles in the deep mining population according to the deep mining learning strategy position update formula;

[0045] Update the positions of particles in the global search population according to the global exploration learning strategy position update formula;

[0046] A fixed percentage of particles are selected from the deep mining population and the global search population, and a crossover population is generated according to the crossover population position update formula generated by the adaptive Lapp crossover operator;

[0047] Merge the deep mining, global exploration population and cross population to get the merged population;

[0048] Use aggressive greedy updates to merge the global optimal solution and historical individual optimal solutions in the population;

[0049] Adjust the particle ratio between the deep mining population and the global search population according to preset rules;

[0050] When the number of iterations t is greater than or equal to the maximum number of iterations T, the iteration is stopped and the global optimal solution, that is, the optimal model parameters, is output.

[0051] Furthermore, the temperature compensation model in the parameter optimization module is a temperature compensation model based on a convolutional neural network and a dendritic network.

[0052] The beneficial technical effects of the present invention are:

[0053] The present invention proposes a temperature compensation method and system for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD. The present invention first obtains pressure data at different temperatures as a training data set; and normalizes the training data set; uses an improved rime algorithm to optimize the model parameters of the temperature compensation model to obtain the optimal model parameters; the model parameters include the initial learning rate and the maximum number of iterations of the neural network; the temperature compensation model is a temperature compensation model based on a convolutional neural network and a dendritic network; the normalized training data set is input into the temperature compensation model, and the temperature compensation model is trained using the optimal model parameters to obtain a trained temperature compensation model; the pressure data to be measured is input into the trained temperature compensation model for temperature compensation to obtain the corrected pressure data. The present invention uses an improved rime algorithm to optimize the neural network parameter optimization to increase the robustness of the model, and then extracts the local features of the temperature compensation data set through a convolutional neural network, and further constructs features through a dendritic network. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0055] Figure 1 It is a flow chart of a temperature compensation method for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD described in an embodiment of the present invention.

[0056] Figure 2 1 is a diagram showing an example structure of a data acquisition system in an embodiment of the present invention.

[0057] Figure 3 1 is an example diagram of pressure data collected at different temperatures in an embodiment of the present invention.

[0058] Figure 4 It is a schematic diagram of the CNN network and DD network structure in an embodiment of the present invention.

[0059] Figure 5 is a density function curve diagram of the Laplace distribution in an embodiment of the present invention.

[0060] Figure 6 It is a flow chart of optimizing the model parameters of the temperature compensation model by using the improved rime algorithm in an embodiment of the present invention.

[0061] Figure 7is a distribution diagram of the absolute error of the pressure sensor in the embodiment of the present invention. DETAILED DESCRIPTION

[0062] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0063] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, equipment, method or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. In this article, it is to be understood that any number of elements in the drawings is for illustration and not limitation, and any naming is only for distinction and does not have any limiting meaning.

[0064] The present invention provides a temperature compensation method and system for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD. Firstly, the improved rime algorithm is used to optimize the initial parameters of the model. Then, a convolutional neural network (CNN) network is used to extract local features in a data set. The extracted features are then sent to a dendritic network (DD) network for feature construction, thereby further improving the sensitivity of the sensor.

[0065] The embodiment of the present invention proposes a temperature compensation method for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD, such as Figure 1 As shown, the method includes:

[0066] S1. Obtain pressure data at different temperatures as a training data set;

[0067] S2, normalizing the training data set;

[0068] S3. Optimizing the model parameters of the temperature compensation model using the improved rime algorithm to obtain the optimal model parameters; the model parameters include an initial learning rate and a maximum number of iterations; the temperature compensation model is a temperature compensation model based on a convolutional neural network;

[0069] S4, inputting the normalized training data set into the temperature compensation model, and training the temperature compensation model using the optimal model parameters to obtain a trained temperature compensation model;

[0070] S5. Input the pressure data to be measured into the trained temperature compensation model to perform temperature compensation, and obtain corrected pressure data.

[0071] The method starts from S1. In S1, pressure data at different temperatures are obtained as a training data set.

[0072] According to an embodiment of the present invention, a data acquisition system is designed using a silicon piezoresistive pressure sensor, an AD conversion module, a host computer, a DS18B20 temperature sensor and an STM32F407 single-chip microcomputer. Figure 2 As shown, data collection is performed. The designed data acquisition system is placed in a high and low temperature box for a specific pressure test. A total of 11 temperature points with a step length of 10°C are set in the range of -20°C to 80°C. The pressure test begins after each temperature point is kept for 180 minutes. During the test, a total of 11 pressure test points with a step length of 4KPa are set. Each pressure is pressurized for 15 minutes, from 0KPa upward to 40KPa, and then from 40KPa downward to 0KPa. The data collected upward and downward are merged in the form of an average, and the results are shown in the figure. Figure 3 As shown. Figure 3 It can be concluded that when the ambient temperature changes, the pressure sensor will show obvious temperature drift effect. The pressure data measured at -20℃~10℃ and 50℃~80℃ are used as the training data set of the model; the pressure data measured at 20℃~40℃ is used as the test data for testing.

[0073] Then, S2 is executed to normalize the training data set.

[0074] According to an embodiment of the present invention, when a data set contains multiple features, if the dimensions of these features are different (for example, length is expressed in meters and centimeters), normalization can help eliminate this dimensional difference, making the different features comparable. Therefore, the pressure data is normalized to eliminate the dimensional effects between the data features. This helps to ensure that each feature contributes relatively evenly to the model, thereby improving the stability and accuracy of the training.

[0075] Normalization can also speed up the convergence of the model. By adjusting the scale of features to a similar range, optimization algorithms such as gradient descent have a more direct path in finding the best solution, thereby speeding up convergence and shortening training time.

[0076] Normalization can also reduce scale sensitivity and improve the accuracy of the algorithm.

[0077] Normalization also helps prevent the gradient explosion problem, that is, during the gradient descent process, the gradient is too large due to the large values ​​of some features, which affects the stability and convergence of the model.

[0078] Normalization can also ensure that each feature has a relatively equal impact on the model, preventing some features from being overemphasized in the model due to their large scale.

[0079] Then, S3 is executed to optimize the model parameters of the temperature compensation model using the improved rime algorithm to obtain the optimal model parameters; the model parameters include an initial learning rate and a maximum number of iterations.

[0080] According to an embodiment of the present invention, the temperature compensation model adopts a temperature compensation model based on a convolutional neural network, or a temperature compensation model based on a convolutional neural network and a dendritic network. The convolutional neural network CNN is a feedforward neural network with local feature extraction capabilities, which are widely used in various fields. The dendritic network DD is a new machine learning algorithm similar to a support vector machine or a multilayer perceptron, and has higher accuracy and model generalization capabilities than the BP neural network.

[0081] like Figure 4 As shown in the figure, the CNN network mainly has two layers of convolution and two layers of pooling. The convolution kernel size in the first and second convolution layers is set to 1×1, including 32 and 64 convolution kernels, and the activation function uses the ReLU function. Both pooling layers use the maximum pooling operation. The DD network mainly has one input layer and a DD layer. The number of network layers in the DD layer is set to 3.

[0082] The key to establishing a good temperature compensation model for silicon piezoresistive pressure sensors is to determine the maximum number of iterations and the initial learning rate of the CNN-DD network. The present invention combines the improved rime algorithm to search for the optimal network parameters of the CNN-DD network, so that the average absolute error of the model in the test set is minimized, thereby improving the model accuracy.

[0083] The rime optimization algorithm (RIME) is a swarm intelligence optimization algorithm based on the physical phenomenon of rime-ice. The algorithm simulates the growth process of soft rime and hard rime of rime-ice, constructs a soft rime search strategy, a hard rime puncture mechanism and an aggressive greedy selection mechanism. The position update formulas of the soft rime search strategy and the hard rime puncture mechanism are shown in equations (1) and (2).

[0084]

[0085] In the formula, Update the position of the particle; R best,j is the best particle. r1 is a random number in the range of (-1,1). w is a constant with a value of 5. ij and Lb ijare the upper and lower bounds of the escape space, respectively. h is the stickiness, ranging from (0,1). t is the current iteration number. r2 is a random number ranging from (0,1). T is the maximum iteration number.

[0086]

[0087] Where r3 is a random number in the range of (-1,1). normr (S i ) is the normalized value of the current particle fitness value.

[0088] For the RIME algorithm, its optimization ability mainly depends on how to balance the algorithm to effectively take into account the global exploration and deep mining capabilities, and how to ensure the diversity of the population. In view of the above problems, the present invention proposes an improved RIME algorithm (IRIME) based on classification discussion and adaptive Laplace crossover operator strategy. The improved rime algorithm mainly redefines the position update formula by using classification discussion and adaptive Laplace crossover operator strategy, balances the RIME algorithm's deep mining and global search capabilities, and increases the diversity of the population, thereby enhancing the algorithm's optimization ability.

[0089] In order to effectively balance global exploration and deep mining, inspired by the idea of ​​classification discussion, combined with the actual situation and the Euclidean distance between the fitness of the previous particle and the optimal particle of the population, the original population is divided into two categories, the deep mining population and the global exploration population, at a ratio of 1:9. As the number of iterations increases, the number of particles in the deep mining population gradually increases, and the number of particles in the global exploration population gradually decreases. Afterwards, a learning strategy focusing on mining is designed for the deep mining population, and a learning strategy tending to global exploration is designed for the global exploration population.

[0090] The deep mining learning strategy position update formula is shown in formula (3):

[0091]

[0092] Where trnd(T+t) is a random number that follows Student's t-distribution. r4 is a random number in the range (0,1). Represents the position of particles in the deep mining population after update.

[0093] The global exploration learning strategy position update formula is shown in formula (4):

[0094]

[0095] In the formula, Indicates the position of particles in the global exploration population after the update. In order to avoid the risk of reducing population diversity due to the clustering strategy, a reasonable population information interaction strategy is particularly important. Experiments have shown that during the convergence process of the IRIME algorithm, some particles will fall into the local optimal problem as the number of iterations increases. In view of this feature, the present invention introduces an adaptive Laplace crossover operator to help the algorithm increase population diversity.

[0096] The density function of Laplace distribution is shown in formula (5):

[0097]

[0098] Among them, x is the value of the random variable; a is the location parameter, and b is the scale parameter. Figure 5 The density function curve of the Laplace distribution is shown.

[0099] If u is a random number with a value range of [0,1] and uniform distribution, the generated random number l is formula (6):

[0100]

[0101] Depend on Figure 5 It can be seen that when b = 1, due to the wide distribution range of f(x), the particles can explore the optimal solution with a larger step size in the early stage of deep mining, avoiding the algorithm from falling into the local optimal problem, as shown in formula (7):

[0102]

[0103] When b = 0.5, since the middle part of f(x) is relatively high, the particle can achieve deep mining with a smaller step in the later stage and can better converge to the global optimal solution, as shown in formula (8):

[0104]

[0105] Where r is a random number in the range [0,1].

[0106] Finally, the crossover population position update formula generated by the adaptive Lapp crossover operator is shown in formula (9):

[0107]

[0108] In the formula is the individual historical optimal solution, r5 is a random number in the range of [0,1], and c is a random number in the range of [-1,1].

[0109] The improved rime algorithm is used to optimize the model parameters of the temperature compensation model. The process of obtaining the optimal model parameters is as follows: Figure 6 As shown, the following steps are included:

[0110] S31, initializing the population; randomly generating multiple groups (e.g., 30 groups) of initial hyperparameter combinations (i.e., individuals in the population), wherein the hyperparameters are the initial learning rate of the CNN network or the CNN-DD network and the maximum number of iterations of the network;

[0111] S32, initialize the fitness function according to formula (10):

[0112]

[0113] In the formula, is the true value of the test sample, y n is the output of the model, and N is the number of test samples;

[0114] S33, initialize the global optimal solution of the particle, the historical individual optimal solution, the maximum number of iterations T (the maximum number of iterations is the number of iterations set in the rime algorithm) and the current number of iterations t=0;

[0115] S34, according to the fitness function of each particle in the population (Formula (10)), the population is divided into a deep mining population and a global search population; for example, there are 10 particles, and according to Formula (10), 10 particle fitness values ​​will be obtained, and then the fitness values ​​are arranged from small to large, and the population is divided according to the proportion, and the population can be divided into a deep mining population and a global search population at a ratio of 1:9;

[0116] S35, updating the position of particles in the deep mining population according to formula (3);

[0117] S36, updating the position of particles in the global search population according to formula (4);

[0118] S37, selecting a fixed percentage (e.g., 5%) of particles from the deep mining population and the global search population to generate a crossover population using formula (9);

[0119] S38, merging the deep mining, global exploration population and cross population to obtain a merged population;

[0120] S39, use aggressive greedy update to merge the global optimal solution and historical individual optimal solution in the population;

[0121] S40, adjusting the particle ratio between the deep mining population and the global search population according to a preset rule; the preset rule is, for example:

[0122]

[0123]

[0124] In the formula, is the proportion of particles in the deep exploration population. is the proportion of particles in the global search population.

[0125] S41, current iteration number t=t+1;

[0126] S42: If the algorithm has not yet reached the stopping criterion (i.e., when t>T), return to S34 and continue iterating; once the stopping condition is reached, proceed to S43.

[0127] S43. Obtain the global optimal solution, that is, the optimal initial learning rate and maximum number of iterations of the CNN-DD network.

[0128] Then, S4 is executed to input the normalized training data set into the temperature compensation model, and the temperature compensation model is trained using the optimal model parameters to obtain a trained temperature compensation model.

[0129] Finally, in S5, the pressure data to be measured is input into the trained temperature compensation model for temperature compensation to obtain corrected pressure data.

[0130] According to an embodiment of the present invention, the pressure data to be measured needs to be normalized first and then input into a trained temperature compensation model, and the model output data needs to be denormalized.

[0131] Furthermore, in order to verify the superiority of the IRIME-CNN-DD model and the necessity of local feature extraction of the CNN network and IRIME optimization network, comparative experiments are used to illustrate the technical effects of the present invention.

[0132] The IRIME-CNN-DD model, RIME-CNN-DD model, CNN-DD model and DD model proposed in the present invention are respectively used for temperature compensation of silicon pressure sensor. Considering the real-time performance of the algorithm, the maximum number of iterations of the IRIME and RIME algorithms in the optimization module is set to 20, and the population size is 30. The results are shown in Table 1.

[0133] Table 1 Model comparison results

[0134]

[0135] The following conclusions can be drawn from Table 1: (1) Compared with DD, the root mean square error of CNN-DD is reduced by 0.20792KPa, and the mean absolute error is reduced by 0.04572KPa. This shows that the local feature extraction of the CNN network can improve the prediction accuracy of the model. (2) Compared with CNN-DD, the root mean square error of RIME-CNN-DD is reduced by 0.06213KPa, and the mean absolute error is reduced by 0.01984KPa. This shows that the RIME algorithm can improve the robustness of the model. (3) Compared with RIME-CNN-DD, the root mean square error of IRIME-CNN-DD is reduced by 0.03626KPa, and the mean absolute error is reduced by 0.02881KPa. This shows that the IRIME algorithm can effectively avoid the local optimal problem of the RIME algorithm.

[0136] In order to further analyze and evaluate the temperature compensation performance of the model, four types of classic neural network temperature compensation models were selected. The optimal network temperature compensation model in 100 independent runs was used as the test model for comparison. The results are shown in Table 2.

[0137] Table 2 Model comparison results

[0138]

[0139] It can be seen from Table 2 that the IRIME-CNN-DD temperature compensation model has the best compensation effect. Compared with the PSO-CNN-DD model, the root mean square error of the model proposed by the present invention is reduced by 30.46%, and the average absolute error is reduced by 32.89%. Compared with the DE-SVM model, the root mean square error of the model proposed by the present invention is reduced by 37.54%, and the average absolute error is reduced by 35.28%. Compared with the GA-BP model, the root mean square error of the model proposed by the present invention is reduced by 41.38%, and the average absolute error is reduced by 38.69%. It can be seen that the two indicators of the model proposed by the present invention are better than traditional silicon piezoresistive pressure sensor temperature compensation models such as PSO-CNN-DD, DE-SVM and GA-BP.

[0140] After the trained model is saved, the measured pressure data is used to obtain the temperature compensation result, and the absolute error is obtained from the temperature compensation result. The calibration pressures are 12KPa, 24KPa, 36KPa, and 40KPa, and the calibration temperature is -20 to 80°C. The temperature compensation results are shown in Table 3, and the absolute error distribution is shown in Figure 7 .

[0141] Table 3 Temperature compensation results

[0142]

[0143] From Table 3 and Figure 7It can be seen that the deviation between the compensation result of the model and the actual result is smaller, which shows that the temperature compensation effect of the silicon piezoresistive pressure sensor designed in the present invention is better, and the output accuracy and reliability of the model meet the requirements of temperature compensation of the silicon piezoresistive pressure sensor.

[0144] Another embodiment of the present invention provides a temperature compensation system for a silicon piezoresistive pressure sensor based on IRIME-CNN-DD, the system comprising:

[0145] A data acquisition module, configured to acquire pressure data at different temperatures as a training data set;

[0146] A preprocessing module, configured to perform normalization processing on the training data set;

[0147] A parameter optimization module, configured to optimize the model parameters of the temperature compensation model using an improved rime algorithm to obtain optimal model parameters; the model parameters include an initial learning rate and a maximum number of iterations; the temperature compensation model is a temperature compensation model based on a convolutional neural network;

[0148] A model training module, configured to input the normalized training data set into the temperature compensation model, and train the temperature compensation model using optimal model parameters to obtain a trained temperature compensation model;

[0149] The temperature compensation module is configured to input the pressure data to be measured into a trained temperature compensation model for temperature compensation to obtain corrected pressure data.

[0150] In this embodiment, optionally, the parameter optimization module uses an improved rime algorithm to optimize the model parameters of the temperature compensation model, and obtaining the optimal model parameters includes:

[0151] Initialize the population and divide it into a deep mining population and a global search population in proportion;

[0152] Initialize the population, fitness function, the global optimal solution of the particle, the historical individual optimal solution of the particle, the maximum number of iterations T and the current number of iterations t;

[0153] According to the fitness function of each particle in the population, the population is divided into a deep mining population and a global search population;

[0154] Iterate the following process:

[0155] Update the positions of particles in the deep mining population according to the deep mining learning strategy position update formula;

[0156] Update the positions of particles in the global search population according to the global exploration learning strategy position update formula;

[0157] A fixed percentage of particles are selected from the deep mining population and the global search population, and a crossover population is generated according to the crossover population position update formula generated by the adaptive Lapp crossover operator;

[0158] Merge the deep mining, global exploration population and cross population to get the merged population;

[0159] Use aggressive greedy updates to merge the global optimal solution and the historical individual optimal solution in the population;

[0160] Adjust the particle ratio between the deep mining population and the global search population according to preset rules;

[0161] When the number of iterations t is greater than or equal to the maximum number of iterations T, the iteration is stopped and the global optimal solution is output, that is, the optimal initial learning rate and maximum number of iterations of the CNN-DD network.

[0162] In this embodiment, optionally, the temperature compensation model in the parameter optimization module is a temperature compensation model based on a convolutional neural network and a dendritic network.

[0163] The function of a silicon piezoresistive pressure sensor temperature compensation system based on IRIME-CNN-DD in an embodiment of the present invention can be described by the aforementioned silicon piezoresistive pressure sensor temperature compensation method based on IRIME-CNN-DD. Therefore, for the parts not described in detail in the system embodiment, please refer to the above method embodiment and will not be repeated here.

[0164] It should be noted that, although several units, modules or submodules are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided into being embodied by multiple modules.

[0165] In addition, although the operations of the method of the present invention are described in a specific order in the drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0166] Although the spirit and principle of the present invention have been described with reference to several specific embodiments, it should be understood that the present invention is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined to benefit, and such division is only for the convenience of expression. The present invention is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims.

Claims

1. A temperature compensation method for silicon piezoresistive pressure sensor based on IRIME-CNN-DD, characterized in that: include: Obtain pressure data at different temperatures as a training data set; Normalizing the training data set; The improved rime algorithm is used to optimize the model parameters of the temperature compensation model to obtain the optimal model parameters; the model parameters include an initial learning rate and a maximum number of iterations; The temperature compensation model is a temperature compensation model based on a convolutional neural network; Inputting the normalized training data set into the temperature compensation model, and training the temperature compensation model using the optimal model parameters to obtain a trained temperature compensation model; The pressure data to be measured is input into the trained temperature compensation model for temperature compensation to obtain the corrected pressure data.

2. The temperature compensation method of silicon piezoresistive pressure sensor based on IRIME-CNN-DD according to claim 1 is characterized in that: The method of optimizing the model parameters of the temperature compensation model by using the improved rime algorithm to obtain the optimal model parameters includes: Initialize the population, fitness function, the global optimal solution of the particle, the historical individual optimal solution of the particle, the maximum number of iterations T and the current number of iterations t; According to the fitness function of each particle in the population, the population is divided into a deep mining population and a global search population; Iterate the following process: Update the positions of particles in the deep mining population according to the deep mining learning strategy position update formula; Update the positions of particles in the global search population according to the global exploration learning strategy position update formula; A fixed percentage of particles are selected from the deep mining population and the global search population, and a crossover population is generated according to the crossover population position update formula generated by the adaptive Lapp crossover operator; Merge the deep mining, global exploration population and cross population to get the merged population; Use aggressive greedy updates to merge the global optimal solution and historical individual optimal solutions in the population; Adjust the particle ratio between the deep mining population and the global search population according to preset rules; When the number of iterations t is greater than or equal to the maximum number of iterations T, the iteration is stopped and the global optimal solution, that is, the optimal model parameters, is output.

3. The temperature compensation method of silicon piezoresistive pressure sensor based on IRIME-CNN-DD according to claim 2 is characterized in that: The deep mining learning strategy position update formula is: In the formula, Represents the position of particles in the deep mining population after update; R best,j represents the best particle; r1 is a random number in the range of (-1,1); h represents the adhesion; Ub ij and Lb ij are the upper and lower bounds of the escape space respectively; r2 is a random number in the range of (0,1); trnd(T+t) is a random number that obeys the Student's t-distribution; r4 is a random number in the range of (0,1).

4. The temperature compensation method of silicon piezoresistive pressure sensor based on IRIME-CNN-DD according to claim 3 is characterized in that: The global exploration learning strategy position update formula is: In the formula, Represents the position of particles in the global exploration population after update.

5. The temperature compensation method of silicon piezoresistive pressure sensor based on IRIME-CNN-DD according to claim 4 is characterized in that: The crossover population position update formula generated by the adaptive Lapp crossover operator is: In the formula, is the individual historical optimal solution; c is a random number in the range of [-1,1]; r5 is a random number in the range of [0,1]; r is a random number in the range [0,1], a is the position parameter, and u is a random number in the range [0,1] that follows a uniform distribution.

6. The temperature compensation method of silicon piezoresistive pressure sensor based on IRIME-CNN-DD according to claim 2 is characterized in that: The fitness function is the mean absolute error.

7. The temperature compensation method of silicon piezoresistive pressure sensor based on IRIME-CNN-DD according to claim 1, characterized in that: The temperature compensation model is a temperature compensation model based on a convolutional neural network and a dendritic network.

8. A temperature compensation system for silicon piezoresistive pressure sensor based on IRIME-CNN-DD, characterized in that: include: A data acquisition module configured to acquire pressure data at different temperatures as a training data set; A preprocessing module, configured to perform normalization processing on the training data set; A parameter optimization module, configured to optimize the model parameters of the temperature compensation model using an improved rime algorithm to obtain optimal model parameters; the model parameters include an initial learning rate and a maximum number of iterations; the temperature compensation model is a temperature compensation model based on a convolutional neural network; A model training module, configured to input the normalized training data set into the temperature compensation model, and train the temperature compensation model using optimal model parameters to obtain a trained temperature compensation model; The temperature compensation module is configured to input the pressure data to be measured into a trained temperature compensation model for temperature compensation to obtain corrected pressure data.

9. The IRIME-CNN-DD based silicon piezoresistive pressure sensor temperature compensation system according to claim 8, characterized in that: The parameter optimization module uses the improved rime algorithm to optimize the model parameters of the temperature compensation model, and obtaining the optimal model parameters includes: Initialize the population, fitness function, the global optimal solution of the particle, the historical individual optimal solution of the particle, the maximum number of iterations T and the current number of iterations t; According to the fitness function of each particle in the population, the population is divided into a deep mining population and a global search population; Iterate the following process: Update the positions of particles in the deep mining population according to the deep mining learning strategy position update formula; Update the positions of particles in the global search population according to the global exploration learning strategy position update formula; A fixed percentage of particles are selected from the deep mining population and the global search population, and a crossover population is generated according to the crossover population position update formula generated by the adaptive Lapp crossover operator; Merge the deep mining, global exploration population and cross population to get the merged population; Use aggressive greedy updates to merge the global optimal solution and historical individual optimal solutions in the population; Adjust the particle ratio between the deep mining population and the global search population according to preset rules; When the number of iterations t is greater than or equal to the maximum number of iterations T, the iteration is stopped and the global optimal solution, that is, the optimal model parameters, is output.

10. The IRIME-CNN-DD based silicon piezoresistive pressure sensor temperature compensation system according to claim 8, characterized in that: The temperature compensation model in the parameter optimization module is a temperature compensation model based on a convolutional neural network and a dendritic network.

Images