Intelligent cabin instrument display method and device, terminal equipment and storage medium

Abstract

The invention provides an intelligent cabin instrument display method and device, terminal equipment and a storage medium. The method comprises the following steps: acquiring an instrument data sample; processing the sample through a fractional order activation function, and calculating output vectors of a hidden layer and an output layer in the classifier model to be trained layer by layer; according to the current prediction error of the classifier model and the fractional derivatives of the weights and the biases of the hidden layer and the output layer, respectively calculating the gradients of the weights and the biases of the hidden layer and the output layer; updating the weight and the offset of the model according to the gradient of the weight and the offset; performing iterative training on the classifier model to obtain a target classifier model; and classifying the acquired intelligent cabin instrument data through the target classifier model to obtain an instrument display classification result. According to the method, a fractional order neural network classification algorithm based on gradient flow optimization is adopted, and data changes of different scales can be adapted through fractional order calculus elements, so that the ability of a classifier model to process nonlinear features is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle intelligent cockpit instrument display, and in particular to an intelligent cockpit instrument display method, device, terminal equipment and storage medium. Background Art

[0002] With the development of automobile technology, intelligent cockpit systems have gradually become the key to improving driving experience and safety. However, existing vehicle intelligent cockpit systems often rely on fixed data collection and display logic, lacking the ability to perceive and respond to changes in the vehicle's internal and external environments in real time, resulting in the instrument display content being not rich and accurate enough, and may not match actual driving needs. In addition, vehicle intelligent cockpit systems usually use traditional data processing and machine learning algorithms, which may not be flexible and accurate enough when dealing with nonlinear features and complex patterns, and have difficulty effectively identifying and classifying subtle changes in multi-source sensor data.

[0003] Therefore, the existing vehicle intelligent cockpit system instrument display not only lacks adaptability to real-time environmental changes, but the corresponding data processing and classification algorithms also have limitations such as lack of flexibility and accuracy. Summary of the invention

[0004] The main purpose of the present invention is to provide an intelligent cockpit instrument display method, device, terminal equipment and storage medium, aiming to solve the technical problems that the existing vehicle intelligent cockpit system instrument display not only has insufficient adaptability to real-time environmental changes, but also the corresponding data processing and classification algorithms also have limitations of inflexibility and accuracy.

[0005] In a first aspect, the present invention provides a smart cockpit instrument display method, comprising:

[0006] Obtain smart cockpit instrument data samples;

[0007] The smart cockpit instrument data sample is processed by a fractional-order activation function, and the output vector of each hidden layer in the classifier model to be trained is calculated layer by layer until the output vector of the output layer is calculated; wherein the classifier model includes an input layer, multiple hidden layers and an output layer;

[0008] Calculating the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer;

[0009] Updating the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer;

[0010] Iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model;

[0011] The collected smart cockpit instrument data is classified by the target classifier model to obtain a smart cockpit instrument display classification result.

[0012] In a specific embodiment, before the smart cockpit instrument data sample is processed by the fractional-order activation function and the output vector of each hidden layer in the classifier model to be trained is calculated layer by layer until the output vector of the output layer is calculated, the method further includes:

[0013] Initializing the initial weights and biases of the classifier model according to an initialization calculation method;

[0014] Among them, the weight W u and the bias b u The initialization calculation methods are expressed as follows:

[0015]

[0016] In the formula, is the initial weight, is the standardized normal distribution, σ is the standard deviation; is the initial bias.

[0017] In a specific embodiment, the calculation formula of the output vector of the hidden layer and the output layer is expressed as:

[0018]

[0019] Where, X u is the input vector, represents the output of the kth layer, k=2,3,…,L, L is the total number of input layers, hidden layers and output layers of the classifier model; For the input vector X u The result after linear transformation; f (k) () is the fractional order activation function;

[0020] Among them, the input vector X u The linear transformation formula It is expressed as:

[0021]

[0022] In the formula, represents the output of the k-1th layer; and are the weight and bias of the kth layer respectively;

[0023] Among them, the implementation method of the fractional-order activation function is expressed as:

[0024]

[0025] In the formula, γ u is a fractional order parameter.

[0026] In a specific embodiment, the step of calculating the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer comprises:

[0027] Compare the output vector of the output layer with the true label to obtain the error currently predicted by the classifier model;

[0028] Back propagation is performed by the chain rule, and the gradients of the weights and biases of the hidden layer and the output layer are calculated respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer.

[0029] In a specific embodiment, comparing the output vector of the output layer with the true label to obtain the error currently predicted by the classifier model includes:

[0030] Compare the output vector of the output layer with the true label, and calculate the error of the current prediction of the classifier model through the cross entropy loss function;

[0031] Among them, the cross entropy loss function is expressed as:

[0032]

[0033] Where, N is the number of smart cockpit instrument data samples; C is the number of categories of smart cockpit instrument data samples; y ij is the true label of the true category of the i-th sample in the j-th category; Predicting the probability that the i-th sample belongs to the j-th class for the classifier model;

[0034] Among them, the probability predicted by the classifier model is The calculation method is expressed as:

[0035]

[0036] In the formula, represents the output of the last layer, and index j represents the score of a specific category.

[0037] In a specific embodiment, the calculation formulas of the gradients of the weights and biases of the hidden layer and the output layer are respectively expressed as:

[0038]

[0039] In the formula, α u is the learning rate, and are the cross entropy loss function L u For the weight W u and the bias b u The gradient of represents the fractional derivative applied to the gradient, γ u is a fractional order parameter;

[0040] The calculation method of the fractional derivative is expressed as follows:

[0041]

[0042] In the formula, and are the cross entropy loss function L u For the weight W u and the bias b u The gradient of u is a fractional order parameter.

[0043] In a specific embodiment, before updating the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer, the method further comprises:

[0044] According to the adaptive gradient synchronization mechanism, the learning rate of each step is adjusted in real time;

[0045] Wherein, the adjustment formula of the learning rate is expressed as:

[0046]

[0047] In the formula, is the adjusted learning rate, is the learning rate before adjustment, λ u is the adjustment factor, is the cross entropy loss function gradient.

[0048] In a second aspect, the present invention provides an intelligent cockpit instrument display device, comprising:

[0049] An instrument data sample acquisition module is used to obtain data samples of smart cockpit instruments;

[0050] An output vector calculation module, used to process the smart cockpit instrument data sample through a fractional-order activation function, and calculate the output vector of each hidden layer in the classifier model to be trained layer by layer until the output vector of the output layer is calculated;

[0051] A weight and bias gradient calculation module, used to calculate the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer;

[0052] A weight and bias update module, used to update the weight and bias of the classifier model according to the gradient of the weight and bias of the hidden layer and the output layer;

[0053] An iterative training module, used for iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model;

[0054] The instrument display classification module is used to classify the collected smart cockpit instrument data through the target classifier model to obtain the smart cockpit instrument display classification result.

[0055] In a third aspect, the present invention provides a terminal device, comprising: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the smart cockpit instrument display method as described in the first aspect is implemented.

[0056] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the smart cockpit instrument display method as described in the first aspect.

[0057] Compared with the prior art, the beneficial effect of the present invention lies in: processing the smart cockpit instrument data samples through a fractional-order activation function, calculating the output vector of each hidden layer in the classifier model to be trained layer by layer, until the output vector of the output layer is calculated, and then calculating the gradients of the weights and biases of the hidden layer and the output layer respectively according to the current predicted error of the classifier model and the fractional-order derivatives of the weights and biases of the hidden layer and the output layer, and then updating the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer. The present invention adopts a fractional-order neural network classification algorithm based on gradient flow optimization, which can adapt to data changes of different scales through fractional-order calculus elements, thereby enhancing the ability of the classifier model to process nonlinear features, and the fractional-order activation function improves the classifier model's ability to recognize complex patterns, making the classification results more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flowchart of a smart cockpit instrument display method provided by one embodiment of the present invention;

[0059] Figure 2 It is a flow chart of a smart cockpit instrument display device provided by one embodiment of the present invention;

[0060] Figure 3 It is a structural diagram of a terminal device provided by an embodiment of the present invention.

[0061] in:

[0062] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0063] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0064] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0065] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0066] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature.

[0067] See also Figure 1 , Figure 1 The present invention is a flowchart of a smart cockpit instrument display method provided by an embodiment of the present invention.

[0068] A smart cockpit instrument display method according to an embodiment of the present invention comprises the following steps:

[0069] S100, obtaining smart cockpit instrument data samples;

[0070] In this embodiment, the smart cockpit instrument data samples are multi-dimensional data inside and outside the vehicle, including the real-time status of the vehicle and external environment information, such as speed, temperature, humidity, noise, light intensity, etc. The data is collected from multi-source sensors on the vehicle, including but not limited to speed sensors, temperature sensors, humidity sensors, sound sensors, and light sensors, so as to fully capture the real-time status of the vehicle and external environment information. The data is stored in structured JSON format.

[0071] In one embodiment, the attributes of the data include:

[0072] R a1 is the speed, in km / h, indicating the current speed of the vehicle;

[0073] R a2 is the external temperature, in °C, indicating the vehicle's external ambient temperature;

[0074] R a3 is the internal temperature, in °C, indicating the vehicle's internal ambient temperature;

[0075] R a4 is humidity, in %, indicating the humidity inside the vehicle;

[0076] R a5 is the noise inside the vehicle, in dB, indicating the noise level inside the vehicle;

[0077] R a6 is the light intensity, in lux, which indicates the light intensity inside the vehicle;

[0078] R a7 It is the GPS coordinate in the format of longitude and latitude, indicating the real-time geographic location of the vehicle;

[0079] R a8 is the engine status, expressed as a binary code, indicating the current working status of the engine;

[0080] R a9 The fuel quantity is in liters, indicating the remaining fuel quantity;

[0081] R a10 It is the status of the vehicle's doors and windows, expressed as a binary code, showing the open and closed status of each door and window of the vehicle.

[0082] It should be noted that this embodiment is only used to illustrate one data format and type of the present invention. In actual applications, the attributes of data are usually more than 10 attributes, and the number of attributes of data may reach dozens or even hundreds.

[0083] Furthermore, the collected data is labeled. The labeling method in this embodiment is manual labeling. In one embodiment, the labeled categories are classified according to the requirements of the vehicle instrument display, such as speed display, temperature display, weather display, etc.

[0084] The present invention uses the collected smart cockpit instrument data samples for classifier model training. Specifically, a fractional-order neural network classification algorithm based on gradient flow optimization is used as the classification algorithm to train the classifier model to be trained for the input smart cockpit instrument data samples. By adopting fractional-order calculus elements to adapt to data changes of different scales, the ability of the classifier model to process nonlinear features is enhanced, thereby realizing the classification of the display types of the smart cockpit instruments.

[0085] The classifier model includes an input layer, multiple hidden layers and an output layer. Each layer of the present invention adopts a fractional activation function to enhance the ability of the classifier model in processing nonlinear features. The number of layers of the classifier model neural network is defined as L, where L is the total number of input layers, hidden layers and output layers of the classifier model, and L≥4, for example, L=4, indicating that the classifier model has one input layer, two hidden layers and one output layer. and are the weight and bias of the kth layer respectively, k=2,3,…,L.

[0086] Specifically, the training process of the fractional-order neural network classification algorithm based on gradient flow optimization of the classifier model is described by the following steps.

[0087] In a specific embodiment, in step S200, the smart cockpit instrument data sample is processed by a fractional-order activation function, and the output vector of each hidden layer in the classifier model to be trained is calculated layer by layer, until the output vector of the output layer is calculated, the method further includes the following steps:

[0088] S700, initializing the initial weights and biases of the classifier model according to an initialization calculation method;

[0089] Among them, the weight W u and the bias b u The initialization calculation methods are expressed as follows:

[0090]

[0091] In the formula, is the initial weight, is the standardized normal distribution, σ is the standard deviation; is the initial bias.

[0092] In this embodiment, the initial weight W is set u and the bias term b u , use small random numbers to initialize, set the starting state for the fractional-order neural network, the initial weight Set to follow a standardized normal distribution and multiply by the standard deviation σ, the initial bias Set to zero vector.

[0093] S200, processing the smart cockpit instrument data sample by a fractional-order activation function, and calculating the output vector of each hidden layer in the classifier model to be trained layer by layer until the output vector of the output layer is calculated.

[0094] In this embodiment, after the smart cockpit instrument data sample is input, it is processed by a fractional-order activation function and calculated layer by layer until the output layer. At each layer, the transformation of the data is controlled by a fractional-order nonlinear function, which enhances the classifier model's ability to recognize complex patterns and makes the classification results more accurate.

[0095] In a specific embodiment, the calculation formula of the output vector of the hidden layer and the output layer is expressed as:

[0096]

[0097] Where, X u is the input vector, represents the output of the kth layer, k=2,3,…,L, L is the total number of input layers, hidden layers and output layers of the classifier model; For the input vector Xu The result after linear transformation; f (k) () is the fractional order activation function;

[0098] Among them, the input vector X u The linear transformation formula It is expressed as:

[0099]

[0100] In the formula, represents the output of the k-1th layer; and are the weight and bias of the kth layer respectively.

[0101] In this embodiment, the fractional activation function f (k) ()Can improve the ability to handle nonlinear features.

[0102] Among them, the implementation method of the fractional-order activation function is expressed as:

[0103]

[0104] In the formula, γ u It is a fractional order parameter, which is used to adjust the steepness of the function curve.

[0105] S300, respectively calculating the gradients of the weights and biases of the hidden layer and the output layer according to the error currently predicted by the classifier model and the fractional-order derivatives of the weights and biases of the hidden layer and the output layer.

[0106] In this embodiment, not only the current prediction error of the classifier model is used, but also the fractional derivatives of the weights and biases of each hidden layer and output layer are used to improve the traditional gradient calculation of weights and biases. The calculation of the gradient not only depends on the prediction error, but is also affected by the fractional order of the current weight state and bias. This makes it possible to adapt to data changes of different scales and enhance the ability of the classifier model to handle nonlinear features.

[0107] In a specific embodiment, the step S300 includes the following steps:

[0108] S310: Compare the output vector of the output layer with the true label to obtain the error of the current prediction of the classifier model.

[0109] In this embodiment, the output vector of the output layer is compared with the true label to evaluate the error of the current prediction of the output layer, that is, the error of the current prediction of the classifier model, so as to provide a basis for the subsequent gradient adjustment of weights and biases.

[0110] In a specific embodiment, the step S310 compares the output vector of the output layer with the true label to obtain the error currently predicted by the classifier model, including:

[0111] S311, comparing the output vector of the output layer with the true label, and calculating the error of the current prediction of the classifier model through the cross entropy loss function;

[0112] Among them, the cross entropy loss function is expressed as:

[0113]

[0114] Where, N is the number of smart cockpit instrument data samples; C is the number of categories of smart cockpit instrument data samples; y ij is the true label of the true category of the i-th sample in the j-th category; Predicting the probability that the i-th sample belongs to the j-th class for the classifier model;

[0115] Among them, the probability predicted by the classifier model is The calculation method is expressed as:

[0116]

[0117] In the formula, Represents the output of the last layer, and index j represents the score of a specific category. The score output by the last layer is converted into a probability value through the Softmax function.

[0118] In this embodiment, the result of the output layer is compared with the true label, and the cross entropy loss function L is used. u Evaluate and calculate the error of the current prediction to provide a basis for the subsequent gradient adjustment of weights and biases.

[0119] S320, performing back propagation through the chain rule, and calculating the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional-order derivatives of the weights and biases of the hidden layer and the output layer.

[0120] In this embodiment, chain rule is used for back propagation to calculate the gradient of each layer. Fractional calculus is used to improve the traditional gradient calculation so that the gradient calculation depends not only on the error, but also on the fractional order of the current weight state and bias, so that it can adapt to data changes of different scales and enhance the ability of the classifier model to handle nonlinear features.

[0121] In a specific embodiment, the calculation formulas of the gradients of the weights and biases of the hidden layer and the output layer are respectively expressed as:

[0122]

[0123] In the formula, α u is the learning rate, and are the cross entropy loss function L u For the weight W u and the bias b u The gradient of represents the fractional derivative applied to the gradient, γ u is a fractional order parameter used to enhance the adaptability of the classifier model to complex data patterns;

[0124] The calculation method of the fractional derivative is expressed as follows:

[0125]

[0126] In the formula, and are the cross entropy loss function L u For the weight W u and the bias b u The gradient of u is a fractional order parameter.

[0127] In a specific embodiment, before the step S400 updates the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer, the method further includes the following steps:

[0128] S800, according to the adaptive gradient synchronization mechanism, adjust the learning rate of each step in real time;

[0129] Wherein, the adjustment formula of the learning rate is expressed as:

[0130]

[0131] In the formula, is the adjusted learning rate, is the learning rate before adjustment, λ u is the adjustment factor, is the cross entropy loss function gradient.

[0132] In this embodiment, the learning rate is dynamically adjusted according to the fluidity of the gradient and the actual response of the network. Specifically, according to the adaptive gradient synchronization mechanism, the learning rate is dynamically adjusted through the modulus of the gradient of the cross entropy loss function to optimize the rapid convergence and accuracy during the training process, thereby achieving a balance between rapid convergence and high accuracy. This mechanism enables the classifier model to adapt to data changes more flexibly and improves the generalization ability of the classifier model.

[0133] S400: Update the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer.

[0134] In this embodiment, the weights and biases of the classifier model are updated using the optimized gradient to reduce the training error and gradually approach the optimal solution. The calculation methods for updating the weights and biases are respectively expressed as:

[0135]

[0136] In the formula, and are the updated weights and biases respectively, and are the weight and bias before update, ΔW u and Δb u are the gradients of weights and biases of each layer respectively.

[0137] In one embodiment, the weights and biases of the classifier model are updated using the optimized gradient and learning rate.

[0138] S500: Iteratively train the classifier model according to a preset stop iteration condition to obtain a target classifier model.

[0139] The above steps S200 to S400 are iterated repeatedly until the preset stop iteration condition is met, which means that the classifier model training is completed and the optimal model parameters are obtained, thereby obtaining the target classifier model.

[0140] In a specific embodiment, the preset condition for stopping iteration is reaching a preset maximum number of iterations. Further, the preset maximum number of iterations is set to 1000 times.

[0141] S600: Classify the collected smart cockpit instrument data using the target classifier model to obtain a smart cockpit instrument display classification result.

[0142] The vehicle's multi-source sensors collect the vehicle's smart cockpit instrument data, including the vehicle's real-time status and external environment information. It is understandable that the attributes of the vehicle status data sample can cover the newly collected smart cockpit instrument data, for example, the smart cockpit instrument data includes one or more of the vehicle's current speed, the vehicle's external ambient temperature, the vehicle's internal ambient temperature, the vehicle's internal humidity, the vehicle's internal noise level, the vehicle's internal light intensity, the vehicle's real-time geographic location, the engine's current working state, the engine's current working state, and the vehicle's doors and windows open and closed state.

[0143] Use the trained target classifier model to classify the display of the smart cockpit instrument. In one embodiment, the collected original smart cockpit instrument data is input into the trained feature extraction and feature dimension reduction model for feature processing, and further, the processed features are input into the target classifier model for classification to obtain the classification results, and then the smart cockpit instrument is controlled to display the classification results. In this embodiment, the display classification categories include speed display, temperature display, weather display, etc., which can adaptively adjust the display content of the cockpit instrument to match the current driving environment and user needs. This intelligent display method not only makes driving information more intuitive and easy to understand, but also helps to reduce the cognitive burden of drivers, allowing them to focus more on driving itself.

[0144] In summary, an intelligent cockpit instrument display method provided by an embodiment of the present invention processes the intelligent cockpit instrument data sample through a fractional-order activation function, calculates the output vector of each hidden layer in the classifier model to be trained layer by layer, until the output vector of the output layer is calculated, and then calculates the gradients of the weights and biases of the hidden layer and the output layer according to the current predicted error of the classifier model and the fractional-order derivatives of the weights and biases of the hidden layer and the output layer, respectively, and then updates the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer. The present invention adopts a fractional-order neural network classification algorithm based on gradient flow optimization, which can adapt to data changes of different scales through fractional-order calculus elements, thereby enhancing the ability of the classifier model to process nonlinear features, and the fractional-order activation function improves the ability of the classifier model to recognize complex patterns, so that the classification result is more accurate.

[0145] In addition, the present invention also adopts an adaptive gradient synchronization mechanism to optimize the speed and accuracy of the training process by dynamically adjusting the learning rate, achieving a balance between fast convergence and high accuracy. This mechanism enables the classifier model to adapt to data changes more flexibly, improving the generalization ability and long-term stability of the classifier model.

[0146] Furthermore, by collecting multi-dimensional data inside and outside the vehicle, such as speed, temperature, humidity, noise, light intensity, etc., and storing them in a structured JSON format, it is possible to fully capture the real-time status of the vehicle and external environment information, and to adaptively adjust the display content of the cockpit instrument to match the current driving environment and user needs. This intelligent display method not only makes driving information more intuitive and easy to understand, but also helps reduce the cognitive burden of drivers, allowing them to focus more on driving itself.

[0147] See also Figure 2 , Figure 2The figure is a schematic diagram of the structure of an intelligent cockpit instrument display device provided by one embodiment of the present invention.

[0148] An intelligent cockpit instrument display device according to an embodiment of the present invention includes:

[0149] An instrument data sample acquisition module is used to obtain data samples of smart cockpit instruments;

[0150] An output vector calculation module, used to process the smart cockpit instrument data sample through a fractional-order activation function, and calculate the output vector of each hidden layer in the classifier model to be trained layer by layer until the output vector of the output layer is calculated; wherein the classifier model includes an input layer, multiple hidden layers and an output layer;

[0151] A weight and bias gradient calculation module, used to calculate the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer;

[0152] A weight and bias update module, used to update the weight and bias of the classifier model according to the gradient of the weight and bias of the hidden layer and the output layer;

[0153] An iterative training module, used for iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model;

[0154] The instrument display classification module is used to classify the collected smart cockpit instrument data through the target classifier model to obtain the smart cockpit instrument display classification result.

[0155] An intelligent cockpit instrument display device provided in an embodiment of the present invention can execute all steps and functions of an intelligent cockpit instrument display method provided in any of the above embodiments, and the specific functions of the device are not described in detail herein.

[0156] See also Figure 3 , Figure 3 It is a structural diagram of a terminal device provided by an embodiment of the present invention.

[0157] The terminal device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned smart cockpit instrument display method embodiment are implemented, for example: Figure 1 Alternatively, the processor implements the functions of each module in the above-mentioned device embodiments when executing the computer program.

[0158] Exemplarily, the computer program may be divided into one or more modules, and the one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device. For example, the computer program may be divided into several modules, and the specific functions of each module have been described in detail in a smart cockpit instrument display method provided in any of the above embodiments, and the specific functions of the device will not be repeated here.

[0159] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory. Those skilled in the art may understand that the schematic diagram is only an example of a terminal device and does not constitute a limitation on a terminal device. The terminal device may include more or fewer components than shown in the figure, or may combine certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0160] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, and uses various interfaces and lines to connect the various parts of the entire terminal device.

[0161] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the intelligent cockpit instrument display method by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0162] An embodiment of the present invention further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned intelligent cockpit instrument display method is implemented.

[0163] If the module integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0164] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A smart cockpit instrument display method, characterized in that: include: Obtain smart cockpit instrument data samples; The smart cockpit instrument data sample is processed by a fractional-order activation function, and the output vector of each hidden layer in the classifier model to be trained is calculated layer by layer until the output vector of the output layer is calculated; wherein the classifier model includes an input layer, multiple hidden layers and an output layer; Calculating the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer; Updating the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer; Iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model; The collected smart cockpit instrument data is classified by the target classifier model to obtain a smart cockpit instrument display classification result.

2. The intelligent cockpit instrument display method according to claim 1, characterized in that: Before processing the smart cockpit instrument data sample by using a fractional-order activation function and calculating the output vector of each hidden layer in the classifier model to be trained layer by layer until the output vector of the output layer is calculated, the method further includes: Initializing the initial weights and biases of the classifier model according to an initialization calculation method; Among them, the weight W u and the bias b u The initialization calculation methods are expressed as follows: In the formula, is the initial weight, is the standardized normal distribution, σ is the standard deviation; is the initial bias.

3. The intelligent cockpit instrument display method according to claim 1, characterized in that: The calculation formula of the output vector of the hidden layer and the output layer is expressed as: Where, X u is the input vector, represents the output of the kth layer, k=2, 3, ..., L, L is the total number of input layers, hidden layers and output layers of the classifier model; For the input vector X u The result after linear transformation; f (k) () is the fractional order activation function; Among them, the input vector X u The linear transformation formula It is expressed as: In the formula, represents the output of the k-1th layer; and are the weight and bias of the kth layer respectively; Among them, the implementation method of the fractional-order activation function is expressed as: In the formula, γ u is a fractional order parameter.

4. The intelligent cockpit instrument display method according to claim 1, characterized in that: The step of calculating the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer comprises: Compare the output vector of the output layer with the true label to obtain the error currently predicted by the classifier model; Back propagation is performed by the chain rule, and the gradients of the weights and biases of the hidden layer and the output layer are calculated respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer.

5. The intelligent cockpit instrument display method according to claim 4, characterized in that: The step of comparing the output vector of the output layer with the true label to obtain the error currently predicted by the classifier model includes: Compare the output vector of the output layer with the true label, and calculate the error of the current prediction of the classifier model through the cross entropy loss function; Among them, the cross entropy loss function is expressed as: Where, N is the number of smart cockpit instrument data samples; C is the number of categories of smart cockpit instrument data samples; y ij is the true label of the true category of the i-th sample in the j-th category; Predicting the probability that the i-th sample belongs to the j-th class for the classifier model; Among them, the probability predicted by the classifier model is The calculation method is expressed as: In the formula, represents the output of the last layer, and index j represents the score of a specific category.

6. The intelligent cockpit instrument display method according to claim 5, characterized in that: The calculation formulas of the gradients of the weights and biases of the hidden layer and the output layer are respectively expressed as: In the formula, α u is the learning rate, and are the cross entropy loss function L u For the weight W u and the bias b u The gradient of represents the fractional derivative applied to the gradient, γ u is a fractional order parameter; The calculation method of the fractional derivative is expressed as follows: In the formula, and are the cross entropy loss function L u For the weight W u and the bias b u The gradient of u is a fractional order parameter.

7. The intelligent cockpit instrument display method according to claim 6, characterized in that: Before updating the weights and biases of the classifier model according to the gradients of the weights and biases of the hidden layer and the output layer, the method further comprises: According to the adaptive gradient synchronization mechanism, the learning rate of each step is adjusted in real time; Wherein, the adjustment formula of the learning rate is expressed as: In the formula, is the adjusted learning rate, is the learning rate before adjustment, λ u is the adjustment factor, is the cross entropy loss function gradient.

8. An intelligent cockpit instrument display device, characterized in that: include: An instrument data sample acquisition module is used to obtain data samples of smart cockpit instruments; An output vector calculation module, used to process the smart cockpit instrument data sample through a fractional-order activation function, and calculate the output vector of each hidden layer in the classifier model to be trained layer by layer until the output vector of the output layer is calculated; wherein the classifier model includes an input layer, multiple hidden layers and an output layer; A weight and bias gradient calculation module, used to calculate the gradients of the weights and biases of the hidden layer and the output layer respectively according to the error currently predicted by the classifier model and the fractional derivatives of the weights and biases of the hidden layer and the output layer; A weight and bias update module, used to update the weight and bias of the classifier model according to the gradient of the weight and bias of the hidden layer and the output layer; An iterative training module, used for iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model; The instrument display classification module is used to classify the collected smart cockpit instrument data through the target classifier model to obtain the smart cockpit instrument display classification result.

9. A terminal device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the intelligent cockpit instrument display method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the intelligent cockpit instrument display method according to any one of claims 1 to 7.

Images