Vehicle state monitoring method and device based on remote vehicle control

Abstract

The invention provides a vehicle state monitoring method and device based on remote vehicle control. The method comprises the following steps: acquiring a vehicle state data sample; initializing a quantum state of each hidden layer node in the classifier model to be trained, and generating an initial weight and an initial bias of a target hidden layer node; according to the initial weight and a quantum entanglement strategy, generating an adjusted weight; performing forward propagation operation on each vehicle state data sample according to the adjusted weight and initial offset to obtain hidden layer output of each vehicle state data sample in a hidden layer and network output from the hidden layer to an output layer; performing iterative training on the classifier model to obtain a target classifier model; and classifying the vehicle state data of the monitored vehicle through the target classifier model to obtain a vehicle state monitoring result of the monitored vehicle. According to the method, the extreme learning machine algorithm is optimized through the quantum entanglement strategy, so that the training and classification processes of the classifier model are optimized, and the classification capability and stability of the classifier model are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle monitoring, and in particular to a vehicle state monitoring method and device based on remote vehicle control. Background Art

[0002] With the development of intelligent transportation systems, real-time monitoring of vehicle status has become increasingly important. Existing vehicle monitoring systems often rely on data collection and processing methods of traditional machine learning models. These methods may encounter problems such as low accuracy, lack of real-time performance, and low data processing efficiency when processing large amounts of complex and dynamically changing vehicle status data. In addition, as vehicle functions increase, the dimensions and complexity of vehicle status data are also increasing, bringing greater challenges to existing monitoring systems.

[0003] Therefore, existing machine learning models are still unable to effectively process complex vehicle status data, resulting in poor accuracy and real-time performance of vehicle status monitoring. Summary of the invention

[0004] The main purpose of the present invention is to provide a vehicle status monitoring method and device based on remote vehicle control, aiming to solve the technical problem that the existing machine learning model is still unable to effectively process complex vehicle status data, resulting in poor accuracy and real-time performance of vehicle status monitoring.

[0005] In a first aspect, the present invention provides a vehicle state monitoring method based on remote vehicle control, comprising:

[0006] Obtain vehicle status data samples;

[0007] Initialize the quantum state of each hidden layer node in the classifier model to be trained, and generate the initial weight and initial bias of the target hidden layer node; wherein the classifier model includes an input layer, a hidden layer and an output layer, and the hidden layer includes a plurality of hidden layer nodes;

[0008] According to the initial weight of the target hidden layer node and the preset first hidden layer parameter, generating the adjusted weight of the target hidden layer node through a quantum entanglement strategy;

[0009] According to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias, a forward propagation operation is performed on each of the vehicle state data samples to obtain a hidden layer output of each of the vehicle state data samples in the hidden layer and a network output from the hidden layer to 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 vehicle status data of the monitored vehicle is classified by the target classifier model to obtain the vehicle status monitoring result of the monitored vehicle.

[0012] In a specific embodiment, the initialization of the quantum state of each hidden layer node in the classifier model to be trained to generate the initial weight and initial bias of the target hidden layer node includes:

[0013] Initializing the quantum state of each hidden layer node in the classifier model to be trained, and generating a composite quantum state of each hidden layer node;

[0014] The initial weight and initial bias of the target hidden layer node are generated by a quantum state estimation strategy and the composite quantum state of each hidden layer node in the classifier model.

[0015] In a specific embodiment, the initializing the quantum state of each hidden layer node in the classifier model to be trained to generate the composite quantum state of each hidden layer node includes:

[0016] Initialize the quantum state of each hidden layer node in the classifier model to be trained, and generate the composite quantum state of each hidden layer node according to the input feature dimension of the vehicle state data sample, the number of hidden layer nodes required by the classifier model and the calculation formula of the composite quantum state;

[0017] Wherein, the calculation formula of the composite quantum state is expressed as:

[0018]

[0019] In the formula, Ψ u is the composite quantum state, u is the input feature dimension of the vehicle state data sample, K is the number of hidden nodes required by the classifier model, α u,k is the complex amplitude of the kth quantum ground state, φ k is the kth quantum ground state.

[0020] In a specific embodiment, the generating of the initial weight and initial bias of the target hidden layer node by the quantum state estimation strategy and the composite quantum state of each hidden layer node in the classifier model includes:

[0021] Generate an initial weight of a target hidden layer node according to the composite quantum state of each hidden layer node in the classifier model and an initial weight calculation formula of the quantum state estimation strategy;

[0022] Generate an initial bias of a target hidden layer node according to the composite quantum state of each hidden layer node in the classifier model and the initial bias calculation formula of the quantum state estimation strategy;

[0023] The initial weight calculation formula and the initial bias calculation formula are respectively expressed as:

[0024] W u,ij =Re(<ψ u,i |ψ u,j >)

[0025] b u,i =Im(<ψ u,i |ψ u >)

[0026] Where W u,ij is the initial weight between the i-th hidden layer node and the j-th hidden layer node, Re() represents the real part, Im() represents the imaginary part, ψ u,i and ψ u,j are the composite quantum states of the i-th hidden layer node and the j-th hidden layer node, ψ u is the global composite quantum state of the hidden layer, b u,i is the initial bias of the i-th hidden layer node.

[0027] In a specific embodiment, the step of generating the adjusted weight of the target hidden layer node by a quantum entanglement strategy according to the initial weight of the target hidden layer node and the preset first hidden layer parameter includes:

[0028] Generate an adjusted weight of the target hidden layer node according to the initial weight of the target hidden layer node, the preset first hidden layer parameter and the weight adjustment formula of the quantum entanglement strategy;

[0029] Wherein, the weight adjustment formula is expressed as:

[0030] W′ u,ij =W u,ii +η u ·sin(Δθ u,ij )

[0031] Where W′ u,ij is the adjusted weight between the i-th hidden layer node and the j-th hidden layer node, W u,ij is the initial weight between the i-th hidden layer node and the j-th hidden layer node, η u is the learning rate, Δθ u,ij is the phase difference between the composite quantum states of the i-th hidden node and the j-th hidden node;

[0032] Among them, the phase difference Δθ between the composite quantum state of the i-th hidden layer node and the j-th hidden layer node is u,ij The calculation method is expressed as:

[0033] Δθ u,ij =arg(<ψ u,i |ψu,j >)

[0034] In the formula, ψ u,i and ψ u,j are the composite quantum states of the i-th hidden layer node and the j-th hidden layer node respectively.

[0035] In a specific embodiment, the forward propagation operation is performed on each of the vehicle state data samples according to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias to obtain the hidden layer output of each of the vehicle state data samples in the hidden layer and the network output from the hidden layer to the output layer, including:

[0036] According to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias, a forward propagation operation is performed on each of the vehicle state data samples using a quantum activation function and a composite quantum operation, and a hidden layer output of each of the vehicle state data samples in the hidden layer is obtained through a calculation formula for the hidden layer output;

[0037] According to the hidden layer output of the hidden layer and the output weight of the output layer, the network output of each vehicle state data sample from the hidden layer to the output layer is obtained by the calculation formula of the network output;

[0038] The calculation formula of the hidden layer output and the calculation formula of the network output are respectively expressed as:

[0039]

[0040] In the formula, h u,i is the activation value of the i-th hidden layer node, σ u ( ) is the quantum activation function, W′ u,ij is the adjusted weight between the i-th hidden layer node and the j-th hidden layer node, x u,j is the input feature of the vehicle status data sample, N is the total number of vehicle status data samples, b u,i is the initial bias of the i-th hidden layer node, W u,io is the output weight from the i-th hidden layer node to the output node, M is the number of nodes from the i-th hidden layer node to the output node, y u output for the network;

[0041] Among them, the calculation method of the quantum activation function σ is expressed as:

[0042]

[0043] Wherein, i represents the imaginary unit.

[0044] In a specific embodiment, before obtaining each vehicle state data sample from the hidden layer to the output layer through the calculation formula of the network output according to the hidden layer output of the hidden layer and the output weight of the output layer, the method further includes:

[0045] Optimizing the output weight of the output layer by using the quantum gradient descent method, and calculating the updated amount of the output weight;

[0046] Obtaining an updated output weight of the output layer according to the output weight and the update amount of the output weight;

[0047] Among them, the update amount of the output weight ΔW u,io The calculation method is expressed as:

[0048]

[0049] In the formula, γ u is the learning rate, is the loss function, is the partial derivative of the loss function with respect to the output weight of the output layer;

[0050] Among them, the loss function The calculation method is expressed as:

[0051]

[0052] In the formula, |ψ target > is the target quantum state, |ψ output > is the quantum state output by the output layer network;

[0053] The calculation method of the updated output weight of the output layer is expressed as:

[0054]

[0055] In the formula, is the updated output weight of the output layer, W u,io is the output weight from the i-th hidden layer node to the output node, ΔW u,io is the update amount of the output weight.

[0056] In a second aspect, the present invention provides a vehicle state monitoring device based on remote vehicle control, comprising:

[0057] A sample acquisition module, used to acquire vehicle status data samples;

[0058] An initial weight and initial bias generation module, used to initialize the quantum state of each hidden layer node in the classifier model to be trained, and generate the initial weight and initial bias of the target hidden layer node; wherein the classifier model includes an input layer, a hidden layer and an output layer, and the hidden layer includes a plurality of hidden layer nodes;

[0059] A weight adjustment module, used to generate an adjusted weight of the target hidden layer node through a quantum entanglement strategy according to the initial weight of the target hidden layer node and a preset first hidden layer parameter;

[0060] A hidden layer output and network output calculation module, used to perform a forward propagation operation on each of the vehicle state data samples according to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias, to obtain the hidden layer output of each of the vehicle state data samples in the hidden layer and the network output from the hidden layer to the output layer;

[0061] An iterative training module, used for iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model;

[0062] The vehicle status monitoring module is used to classify the collected vehicle status data of the monitored vehicle through the target classifier model to obtain the vehicle status monitoring result of the monitored vehicle.

[0063] 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 vehicle status monitoring method based on remote vehicle control as described in the first aspect is implemented.

[0064] In a fourth aspect, the present invention provides a computer-readable storage medium, which 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 vehicle status monitoring method based on remote vehicle control as described in the first aspect.

[0065] Compared with the prior art, the beneficial effects of the present invention are: optimizing the extreme learning machine algorithm through the quantum entanglement strategy to optimize the training and classification process of the classifier model, and improving the classification ability and stability of the classifier model. At the same time, the characteristics of quantum computing are used to improve the ability of the classifier model to process complex data, thereby achieving more accurate and faster classification of vehicle status data and realizing real-time monitoring of vehicle status. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a flow chart of a vehicle state monitoring method based on remote vehicle control provided by one embodiment of the present invention;

[0067] Figure 2 It is a flow chart of a vehicle state monitoring device based on remote vehicle control provided by one embodiment of the present invention;

[0068] Figure 3 It is a structural diagram of a terminal device provided by an embodiment of the present invention.

[0069] in:

[0070] 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

[0071] 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.

[0072] 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.

[0073] 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.

[0074] 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.

[0075] See also Figure 1 , Figure 1 It is a flow chart of a vehicle status monitoring method based on remote vehicle control provided by one embodiment of the present invention.

[0076] A vehicle state monitoring method based on remote vehicle control according to an embodiment of the present invention comprises the following steps:

[0077] S100, obtaining a vehicle status data sample;

[0078] In this embodiment, the vehicle status data samples are collected from the remote vehicle control system, which mainly collects the vehicle's operating status and environmental feedback information. The data storage format is JSON.

[0079] In one embodiment, the attributes of the vehicle status data sample include:

[0080] Speed ​​V a is the current speed of the vehicle, in km / h;

[0081] Position P a The current geographic coordinates of the vehicle in the format of longitude and latitude;

[0082] Throttle opening T a is the position of the accelerator pedal, expressed in percentage;

[0083] Brake status B a The state of the brake pedal, represented by a Boolean value;

[0084] Steering wheel angle S a is the rotation angle of the steering wheel, in degrees;

[0085] Engine speed E a is the engine speed in rpm;

[0086] Vehicle status C a The current working status of the vehicle, such as starting, shutting down, etc.;

[0087] Ambient temperature Tm a is the external environment temperature, in °C;

[0088] Noise in the car a is the noise level inside the car, in dB;

[0089] Fuel quantity F a The remaining fuel volume is in liters.

[0090] It should be noted that this embodiment is only for illustrating the format and type of the vehicle status data sample of the present invention. In actual applications, the attributes of the vehicle status data sample are usually more than 10 attributes, and the number of attributes of the data may reach dozens or even hundreds, which is not specifically limited here.

[0091] Furthermore, the collected vehicle status data samples are labeled. The labeling method in this embodiment is manual labeling. In one embodiment, the labeled categories include "normal" and "abnormal".

[0092] The present invention uses the collected vehicle status data samples for classifier model training, adopts the extreme learning machine algorithm based on the quantum entanglement strategy as the classification algorithm, trains the classifier model to be trained with the input vehicle status data samples, and uses the quantum entanglement strategy to enhance the classification ability of the classifier model to obtain the target classifier model.

[0093] Among them, Extreme Learning Machine (ELM) is a machine learning method for a single-layer neural network. Its purpose is to speed up the efficiency of machine learning while ensuring accuracy. Its structure includes input layer, hidden layer and output layer. The input layer is responsible for receiving the input sample vector and outputting sample features. The hidden layer includes multiple hidden layer nodes. The output of each hidden layer node is a weighted sum. The number of neurons in the output layer is equal to the number of sample categories. For binary classification problems, the output layer has two neurons, representing positive and negative categories respectively, and the output result is a vector. The core idea is that the input layer weights and hidden layer biases of the single-layer neural network can be directly randomly selected without adjustment, and the optimal model parameters can be obtained by adjusting the output layer weights only.

[0094] Specifically, the training process of the extreme learning machine algorithm based on the quantum entanglement strategy of the classifier model is described by the following steps.

[0095] S200, initializing the quantum state of each hidden layer node in the classifier model to be trained, and generating the initial weight and initial bias of the target hidden layer node.

[0096] In this embodiment, the quantum state of each hidden layer node in the hidden layer of the classifier model to be trained is initialized to generate the initial weight and initial bias of the target hidden layer node. Especially when facing high-dimensional and complex data, the characteristics of quantum computing are used to improve the ability of the classifier model to process complex data, thereby achieving more accurate and faster vehicle status classification and real-time monitoring of vehicle status.

[0097] In a specific embodiment, the step S200 initializes the quantum state of each hidden layer node in the classifier model to be trained, generates the initial weight and initial bias of the target hidden layer node, and includes the following steps:

[0098] S210, initializing the quantum state of each hidden layer node in the classifier model to be trained, and generating a composite quantum state of each hidden layer node;

[0099] S220, generating an initial weight and an initial bias of a target hidden layer node through a quantum state estimation strategy and the composite quantum state of each hidden layer node in the classifier model.

[0100] In this embodiment, the quantum state of each hidden layer node in the hidden layer of the classifier model is initialized to generate an initial quantum state representation, thereby generating a composite quantum state for each hidden layer node, and then the initial weight and initial bias of the target hidden layer node are generated through the quantum state estimation strategy and the composite quantum state of each hidden layer node in the classifier model.

[0101] In a specific embodiment, the step S210 initializes the quantum state of each hidden layer node in the classifier model to be trained, and generates the composite quantum state of each hidden layer node, including the following steps:

[0102] S211, initializing the quantum state of each hidden layer node in the classifier model to be trained, and generating the composite quantum state of each hidden layer node according to the input feature dimension of the vehicle state data sample, the number of hidden layer nodes required by the classifier model and the calculation formula of the composite quantum state;

[0103] Wherein, the calculation formula of the composite quantum state is expressed as:

[0104]

[0105] In the formula, Ψ u is the composite quantum state, u is the input feature dimension of the vehicle state data sample, K is the number of hidden nodes required by the classifier model, α u,k is the complex amplitude of the kth quantum ground state, φ k is the kth quantum ground state.

[0106] In a specific embodiment, the step S220 generates the initial weight and initial bias of the target hidden layer node through the quantum state estimation strategy and the composite quantum state of each hidden layer node in the classifier model, including the following steps:

[0107] S221, generating an initial weight of a target hidden layer node according to the composite quantum state of each hidden layer node in the classifier model and the initial weight calculation formula of the quantum state estimation strategy;

[0108] S222, generating an initial bias of a target hidden layer node according to the composite quantum state of each hidden layer node in the classifier model and the initial bias calculation formula of the quantum state estimation strategy;

[0109] The initial weight calculation formula and the initial bias calculation formula are respectively expressed as:

[0110] W u,ij =Re(<ψ u,i |ψ u,j >)

[0111] b u,i =Im(<ψ u,i |ψ u >)

[0112] Where W u,ij is the initial weight between the i-th hidden layer node and the j-th hidden layer node, Re() represents the real part, Im() represents the imaginary part, ψ u,i and ψ u,j are the composite quantum states of the i-th hidden layer node and the j-th hidden layer node, ψ u is the global composite quantum state of the hidden layer, b u,i is the initial bias of the i-th hidden layer node.

[0113] In this embodiment, the target hidden layer node refers to the i-th hidden layer node, i=1,2,3,…,K.

[0114] According to the calculation formula of the composite quantum state, ψ u,i , u,j , u Respectively expressed as:

[0115]

[0116] In the formula, α u,i is the complex amplitude of the i-th quantum ground state, φ i is the i-th quantum ground state, α u,j is the complex amplitude of the jth quantum ground state, φ j is the jth quantum ground state.

[0117] S300, generating an adjusted weight of the target hidden layer node through a quantum entanglement strategy according to the initial weight of the target hidden layer node and a preset first hidden layer parameter.

[0118] In this embodiment, after the first hidden layer parameters are set, the quantum entanglement strategy is executed to generate the adjusted weights of the target hidden layer nodes. Through the entanglement operation of quantum bits, the connection strength between the hidden layer nodes is dynamically adjusted to enhance the adaptability of the classifier model.

[0119] In a specific embodiment, step S300 includes the following steps:

[0120] S310, generating an adjusted weight of the target hidden layer node according to the initial weight of the target hidden layer node, the preset first hidden layer parameter and the weight adjustment formula of the quantum entanglement strategy;

[0121] Wherein, the weight adjustment formula is expressed as:

[0122] W′ u,ij =W u,ij +η u ·sin(Δθ u,ij )

[0123] Where W′ u,ij is the adjusted weight between the i-th hidden layer node and the j-th hidden layer node, W u,ij is the initial weight between the i-th hidden layer node and the j-th hidden layer node, η u is the learning rate, Δθ u,ij is the phase difference between the composite quantum states of the i-th hidden node and the j-th hidden node;

[0124] Among them, the phase difference Δθ between the composite quantum state of the i-th hidden layer node and the j-th hidden layer node is u,ij The calculation method is expressed as:

[0125] Δθ u,ij =arg(<ψ u,i |ψ u,j >)

[0126] In the formula, ψ u,i and ψ u,h are the composite quantum states of the i-th hidden layer node and the j-th hidden layer node respectively.

[0127] S400, performing a forward propagation operation on each of the vehicle status data samples according to the preset second hidden layer parameters, the adjusted weights of the target hidden layer nodes and the initial bias, to obtain the hidden layer output of each of the vehicle status data samples in the hidden layer and the network output from the hidden layer to the output layer.

[0128] In this embodiment, forward propagation is performed using the set second hidden layer parameters to calculate the expression of each vehicle state data sample in the hidden layer and the subsequent output layer response.

[0129] In a specific embodiment, the forward propagation is performed using quantum activation functions and composite quantum operations, which can improve the computational efficiency of the classifier model.

[0130] Specifically, step S400 includes the following steps:

[0131] S410, according to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias, a forward propagation operation is performed on each of the vehicle state data samples using a quantum activation function and a composite quantum operation, and a hidden layer output of each of the vehicle state data samples in the hidden layer is obtained by a calculation formula of the hidden layer output;

[0132] S420, according to the hidden layer output of the hidden layer and the output weight of the output layer, obtain the network output of each vehicle state data sample from the hidden layer to the output layer through the calculation formula of the network output;

[0133] The calculation formula of the hidden layer output and the calculation formula of the network output are respectively expressed as:

[0134]

[0135] In the formula, h u,i is the activation value of the i-th hidden layer node, σ u () is the quantum activation function, W u,ij is the adjusted weight between the i-th hidden layer node and the j-th hidden layer node, x u,j is the input feature of the vehicle status data sample, N is the total number of vehicle status data samples, b u,i is the initial bias of the i-th hidden layer node, W u,io is the output weight from the i-th hidden layer node to the output node, M is the number of nodes from the i-th hidden layer node to the output node, y u output for the network;

[0136] Wherein, the quantum activation function σ u The calculation method is expressed as:

[0137]

[0138] Wherein, i represents the imaginary unit.

[0139] Among them, the output weight W from the i-th hidden layer node to the output node u,io It is expressed as:

[0140] W u,io =Re(<ψ u,i |ψ u,o >)

[0141] Where W u,io is the output weight between the i-th hidden layer node and the output node, Re() represents the real part, ψ u,i and ψ u,o are the composite quantum states of the i-th hidden layer node and output node respectively.

[0142] In a specific embodiment, in step S420, according to the hidden layer output of the hidden layer and the output weight of the output layer, each of the vehicle state data samples is obtained by the calculation formula of the network output before the network output from the hidden layer to the output layer, and the vehicle state monitoring method based on remote vehicle control further includes the following steps:

[0143] 430. Optimizing the output weight of the output layer by using a quantum gradient descent method, and calculating an updated amount of the output weight;

[0144] 440. Obtain an updated output weight of the output layer according to the output weight and the update amount of the output weight;

[0145] Among them, the update amount of the output weight ΔW u,io The calculation method is expressed as:

[0146]

[0147] In the formula, γ u is the learning rate, is the loss function, is the partial derivative of the loss function with respect to the output weight of the output layer;

[0148] Among them, the loss function The calculation method is expressed as:

[0149]

[0150] In the formula, |ψ target > is the target quantum state, |ψ output > is the quantum state output by the output layer network;

[0151] The calculation method of the updated output weight of the output layer is expressed as:

[0152]

[0153] In the formula, is the updated output weight of the output layer, Wu,io is the output weight from the i-th hidden layer node to the output node, ΔW u,io is the update amount of the output weight.

[0154] In this embodiment, the quantum gradient descent method is used in the output layer to optimize the output weights of the output layer. The calculation of the quantum gradient is not only fast, but also can accurately capture the output weight adjustment direction that leads to the minimum error.

[0155] For the loss function This embodiment adopts the above-mentioned loss function based on quantum distance measurement, which is particularly suitable for processing data represented by quantum states. The loss function measures the degree of overlap between two quantum states. Minimizing the loss function is to maximize the quantum state overlap between the target and the output.

[0156] Furthermore, the calculation method of the updated output weight of the output layer is expressed as:

[0157]

[0158] In the formula, Re() means taking the real part, Represents the output quantum state with respect to the output weight W of the output layer u,io The partial derivative of .

[0159] And, the calculation method of the update amount of the initial bias is expressed as:

[0160]

[0161] In the formula, Im() means taking the imaginary part, is the output quantum state with respect to the initial bias b u,i The partial derivative of .

[0162] Then, the output weight of the output layer after updating is The calculation method is expressed as:

[0163]

[0164] Updated initial bias The calculation method is expressed as:

[0165]

[0166] This embodiment updates the output weights of the output layer and the initial bias of the hidden layer of the classifier model, which can reduce the training error and gradually approach the optimal solution of the loss function.

[0167] S500: Iteratively train the classifier model according to a preset stop iteration condition to obtain a target classifier model.

[0168] 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.

[0169] In a specific embodiment, the preset condition for stopping iteration is reaching a preset maximum number of iterations. Preferably, the preset maximum number of iterations is set to 1000 times.

[0170] S600: classify the collected vehicle status data of the monitored vehicle by using the target classifier model to obtain a vehicle status monitoring result of the monitored vehicle.

[0171] The remote vehicle control system collects vehicle status data of the monitored vehicle, such as the vehicle's operating status and environmental feedback information. It is understandable that the vehicle status data sample may include newly collected vehicle status data of the monitored vehicle, for example, the vehicle status data of the monitored vehicle includes one or more of the vehicle's current speed, the vehicle's current geographic coordinates, the position of the accelerator pedal, the state of the brake pedal, the steering wheel's rotation angle, the engine's speed, the vehicle's current operating status, the external ambient temperature, the noise level inside the vehicle, and the remaining fuel.

[0172] The trained target classifier model is used to process the vehicle status data of the monitored vehicle, and the target classifier model classifies the vehicle status data of the monitored vehicle to obtain the vehicle status monitoring result of the monitored vehicle, thereby realizing real-time monitoring of the vehicle status. In this embodiment, the classification categories include "normal" and "abnormal".

[0173] In summary, the vehicle state monitoring method based on remote vehicle control provided by the embodiment of the present invention optimizes the training and classification process of the classifier model by optimizing the extreme learning machine algorithm through the quantum entanglement strategy, thereby improving the classification ability and stability of the classifier model. At the same time, the characteristics of quantum computing are used to improve the ability of the classifier model to process complex data, thereby achieving more accurate and faster classification of vehicle state data and realizing real-time monitoring of the vehicle state. In addition, the quantum entanglement strategy is used to dynamically adjust the connection strength between hidden layer nodes to enhance the adaptability of the classifier model. Furthermore, quantum activation functions and composite quantum operations are used for forward propagation to improve the computational efficiency of the classifier model.

[0174] See also Figure 2 , Figure 2 It is a structural schematic diagram of a vehicle state monitoring device based on remote vehicle control provided by one embodiment of the present invention.

[0175] A vehicle state monitoring device based on remote vehicle control according to an embodiment of the present invention comprises:

[0176] A sample acquisition module, used to acquire vehicle status data samples;

[0177] An initial weight and initial bias generation module, used to initialize the quantum state of each hidden layer node in the classifier model to be trained, and generate the initial weight and initial bias of the target hidden layer node; wherein the classifier model includes an input layer, a hidden layer and an output layer, and the hidden layer includes a plurality of hidden layer nodes;

[0178] A weight adjustment module, used to generate an adjusted weight of the target hidden layer node through a quantum entanglement strategy according to the initial weight of the target hidden layer node and a preset first hidden layer parameter;

[0179] A hidden layer output and network output calculation module, used to perform a forward propagation operation on each of the vehicle state data samples according to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias, to obtain the hidden layer output of each of the vehicle state data samples in the hidden layer and the network output from the hidden layer to the output layer;

[0180] An iterative training module, used for iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model;

[0181] The vehicle status monitoring module is used to classify the collected vehicle status data of the monitored vehicle through the target classifier model to obtain the vehicle status monitoring result of the monitored vehicle.

[0182] A vehicle state monitoring device based on remote vehicle control provided in an embodiment of the present invention can execute all steps and functions of a vehicle state monitoring method based on remote vehicle control provided in any of the above embodiments, and the specific functions of the device are not described in detail herein.

[0183] See also Figure 3 , Figure 3 It is a structural diagram of a terminal device provided by an embodiment of the present invention.

[0184] 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 vehicle state monitoring method based on remote vehicle control in the above-mentioned embodiments 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.

[0185] 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 that can perform 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 vehicle status monitoring method based on remote vehicle control provided in any of the above embodiments, and the specific functions of the device will not be repeated here.

[0186] 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.

[0187] 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.

[0188] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the vehicle state monitoring method based on remote vehicle control 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.

[0189] 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 vehicle state monitoring method based on remote vehicle control is implemented.

[0190] 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.

[0191] 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 vehicle state monitoring method based on remote vehicle control, characterized in that: include: Obtain vehicle status data samples; Initialize the quantum state of each hidden layer node in the classifier model to be trained, and generate the initial weight and initial bias of the target hidden layer node; wherein the classifier model includes an input layer, a hidden layer and an output layer, and the hidden layer includes a plurality of hidden layer nodes; According to the initial weight of the target hidden layer node and the preset first hidden layer parameter, generating the adjusted weight of the target hidden layer node through a quantum entanglement strategy; According to the preset second hidden layer parameters, the adjusted weights of the target hidden layer nodes and the initial bias, a forward propagation operation is performed on each of the vehicle state data samples to obtain a hidden layer output of each of the vehicle state data samples in the hidden layer and a network output from the hidden layer to the output layer; Iteratively training the classifier model according to a preset stop iteration condition to obtain a target classifier model; The collected vehicle status data of the monitored vehicle is classified by the target classifier model to obtain the vehicle status monitoring result of the monitored vehicle.

2. The vehicle state monitoring method based on remote vehicle control according to claim 1 is characterized in that: Initializing the quantum state of each hidden layer node in the classifier model to be trained to generate the initial weight and initial bias of the target hidden layer node includes: Initializing the quantum state of each hidden layer node in the classifier model to be trained, and generating a composite quantum state of each hidden layer node; The initial weight and initial bias of the target hidden layer node are generated by a quantum state estimation strategy and the composite quantum state of each hidden layer node in the classifier model.

3. The vehicle state monitoring method based on remote vehicle control according to claim 2 is characterized in that: Initializing the quantum state of each hidden layer node in the classifier model to be trained to generate a composite quantum state of each hidden layer node includes: Initialize the quantum state of each hidden layer node in the classifier model to be trained, and generate the composite quantum state of each hidden layer node according to the input feature dimension of the vehicle state data sample, the number of hidden layer nodes required by the classifier model and the calculation formula of the composite quantum state; Wherein, the calculation formula of the composite quantum state is expressed as: In the formula, Ψ u is the composite quantum state, u is the input feature dimension of the vehicle state data sample, K is the number of hidden nodes required by the classifier model, α u,k is the complex amplitude of the kth quantum ground state, φ k is the kth quantum ground state.

4. The vehicle state monitoring method based on remote vehicle control according to claim 3 is characterized in that: The method of generating the initial weight and initial bias of the target hidden layer node by using the quantum state estimation strategy and the composite quantum state of each hidden layer node in the classifier model includes: Generate an initial weight of a target hidden layer node according to the composite quantum state of each hidden layer node in the classifier model and an initial weight calculation formula of the quantum state estimation strategy; Generate an initial bias of a target hidden layer node according to the composite quantum state of each hidden layer node in the classifier model and the initial bias calculation formula of the quantum state estimation strategy; The initial weight calculation formula and the initial bias calculation formula are respectively expressed as: W u,ij =Re(<ψ u,i |ψ u,j >) b u,i =I(<ψ u,i |ψ u >) Where W u,ij is the initial weight between the i-th hidden layer node and the j-th hidden layer node, Re() represents the real part, Im() represents the imaginary part, ψ u,i and ψ u,j are the composite quantum states of the i-th hidden layer node and the j-th hidden layer node, ψ u is the global composite quantum state of the hidden layer, b u,i is the initial bias of the i-th hidden layer node.

5. The vehicle state monitoring method based on remote vehicle control according to claim 4 is characterized in that: The step of generating the adjusted weight of the target hidden layer node by using a quantum entanglement strategy according to the initial weight of the target hidden layer node and the preset first hidden layer parameter comprises: Generate an adjusted weight of the target hidden layer node according to the initial weight of the target hidden layer node, the preset first hidden layer parameter and the weight adjustment formula of the quantum entanglement strategy; Wherein, the weight adjustment formula is expressed as: W′ u,ij =W u,ij +n u ·sin(Δθ u,ij ) Where W′ u,ij is the adjusted weight between the i-th hidden layer node and the j-th hidden layer node, W u,uj is the initial weight between the i-th hidden layer node and the j-th hidden layer node, η u is the learning rate, Δθ u,ij is the phase difference between the composite quantum states of the i-th hidden node and the j-th hidden node; Among them, the phase difference Δθ between the composite quantum state of the i-th hidden layer node and the j-th hidden layer node is u,ij The calculation method is expressed as: Dth u,ij =arg(<ψ u,i |ψ u,j >) In the formula, ψ u,i and ψ u,j are the composite quantum states of the i-th hidden layer node and the j-th hidden layer node respectively.

6. The vehicle state monitoring method based on remote vehicle control according to claim 5 is characterized in that: The method of performing a forward propagation operation on each of the vehicle state data samples according to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias to obtain the hidden layer output of each of the vehicle state data samples in the hidden layer and the network output from the hidden layer to the output layer includes: According to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias, a forward propagation operation is performed on each of the vehicle state data samples using a quantum activation function and a composite quantum operation, and a hidden layer output of each of the vehicle state data samples in the hidden layer is obtained through a calculation formula for the hidden layer output; According to the hidden layer output of the hidden layer and the output weight of the output layer, the network output of each vehicle state data sample from the hidden layer to the output layer is obtained by the calculation formula of the network output; The calculation formula of the hidden layer output and the calculation formula of the network output are respectively expressed as: In the formula, h u,i is the activation value of the i-th hidden layer node, σ u () is the quantum activation function, W′ u,ij is the adjusted weight between the i-th hidden layer node and the j-th hidden layer node, x u,j is the input feature of the vehicle status data sample, N is the total number of vehicle status data samples, b u,i is the initial bias of the i-th hidden layer node, W u,io is the output weight from the i-th hidden layer node to the output node, M is the number of nodes from the i-th hidden layer node to the output node, y u output for the network; Wherein, the quantum activation function σ u The calculation method is expressed as: Wherein, i represents the imaginary unit.

7. The vehicle state monitoring method based on remote vehicle control according to claim 6 is characterized in that: Before the network output from the hidden layer to the output layer is obtained by the calculation formula of the network output according to the hidden layer output of the hidden layer and the output weight of the output layer, the method further comprises: Optimizing the output weight of the output layer by using the quantum gradient descent method, and calculating the updated amount of the output weight; Obtaining an updated output weight of the output layer according to the output weight and the update amount of the output weight; Among them, the update amount of the output weight ΔW u,io The calculation method is expressed as: In the formula, γ u is the learning rate, is the loss function, is the partial derivative of the loss function with respect to the output weight of the output layer; Among them, the loss function The calculation method is expressed as: In the formula, |ψ target > is the target quantum state, |ψ output > is the quantum state output by the output layer network; The calculation method of the updated output weight of the output layer is expressed as: In the formula, is the updated output weight of the output layer, W u,io is the output weight from the i-th hidden layer node to the output node, ΔW u,io is the update amount of the output weight.

8. A vehicle state monitoring device based on remote vehicle control, characterized in that: include: A sample acquisition module, used to acquire vehicle status data samples; An initial weight and initial bias generation module, used to initialize the quantum state of each hidden layer node in the classifier model to be trained, and generate the initial weight and initial bias of the target hidden layer node; wherein the classifier model includes an input layer, a hidden layer and an output layer, and the hidden layer includes a plurality of hidden layer nodes; A weight adjustment module, used to generate an adjusted weight of the target hidden layer node through a quantum entanglement strategy according to the initial weight of the target hidden layer node and a preset first hidden layer parameter; A hidden layer output and network output calculation module, used to perform a forward propagation operation on each of the vehicle state data samples according to the preset second hidden layer parameters, the adjusted weight of the target hidden layer node and the initial bias, to obtain the hidden layer output of each of the vehicle state data samples in the hidden layer and the network output from the hidden layer to 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 vehicle status monitoring module is used to classify the collected vehicle status data of the monitored vehicle through the target classifier model to obtain the vehicle status monitoring result of the monitored vehicle.

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 when the processor executes the computer program, the vehicle state monitoring method based on remote vehicle control as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the vehicle state monitoring method based on remote vehicle control as described in any one of claims 1 to 7.

Images