Cloud-control integrated intelligent driving automatic collaborative test platform and control method

Abstract

The present application relates to the field of intelligent driving, and specifically to a cloud-controlled integrated intelligent driving automatic collaborative test platform and control method. It uses advanced artificial intelligence technology and natural language processing technology to understand the user's test needs and intentions from the search conditions entered by the user, and matches them with the text semantic features describing the test scenario, so as to return highly matched test scenarios to the user. In this way, fully automated matching of massive scenarios is achieved to meet the diverse needs of different users for testing autonomous driving systems.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving, and specifically to a cloud-controlled integrated intelligent driving automatic collaborative testing platform and control method. Background Art

[0002] With the rapid development of autonomous driving technology, the testing and verification of intelligent driving systems has become particularly important. Traditional testing methods often rely on manually set test scenarios, which is not only inefficient but also difficult to cover all possible traffic environments and driving behaviors. In order to improve the comprehensiveness and automation of testing, test platforms for intelligent driving have emerged.

[0003] In the field of autonomous driving, the diversity and complexity of test scenarios are key factors in ensuring system reliability. However, different users may have different test requirements, such as safety testing, performance testing, or testing of specific functions. Therefore, how to intelligently and accurately screen out scenarios that meet specific test requirements based on the user's test requirements is an important technical issue. Summary of the invention

[0004] This application is made in consideration of the above problems. One purpose of this application is to provide a cloud-controlled integrated intelligent driving automatic collaborative testing platform and control method.

[0005] The embodiment of the present application provides a control method for a cloud-control integrated intelligent driving automatic collaborative test platform, which includes:

[0006] Get the search criteria entered by the user;

[0007] Extracting text descriptions of candidate test scenarios from a test scenario database;

[0008] Performing semantic encoding on the search condition to obtain a search condition semantic encoding feature vector;

[0009] Performing semantic encoding and internal related semantic enhancement on the text description of the candidate test scenario to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors;

[0010] Inputting the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector into a unidirectional attention granularity scanning interactive matching module to obtain a unidirectional matching result representation vector;

[0011] A search result is determined based on the one-way matching result representation vector, where the search result is used to indicate whether to return the candidate test scenario.

[0012] For example, according to the control method of the cloud-control integrated intelligent driving automatic collaborative test platform of an embodiment of the present application, the search condition is semantically encoded to obtain a search condition semantic encoding feature vector, including:

[0013] The search condition is segmented and then input into the Bert model to obtain a sequence of semantic feature vectors of the search condition text;

[0014] The sequence of the search condition text semantic feature vectors is cascaded to obtain the search condition semantic encoding feature vector.

[0015] For example, according to the control method of the cloud-control integrated intelligent driving automatic collaborative test platform of the embodiment of the present application, the text description of the candidate test scenario is semantically encoded and internally related semantically enhanced to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors, including:

[0016] After word segmentation processing is performed on the text description of the candidate test scenario, the text description is input into a test scenario semantic encoder including a Bert model and an LSTM model to obtain a sequence of test scenario word granularity semantic encoding feature vectors;

[0017] The sequence of the test scenario word granularity semantic encoding feature vectors is input into a semantic enhancement module that uses the internal correlation of the sequence to obtain the sequence of enhanced test scenario word granularity semantic encoding feature vectors.

[0018] For example, according to the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform of the embodiment of the present application, the text description of the alternative test scenario is subjected to word segmentation processing and then input into a test scenario semantic encoder including a Bert model and an LSTM model to obtain a sequence of test scenario word granularity semantic encoding feature vectors, including:

[0019] Segmenting the text description of the candidate test scenario to obtain a sequence of text description words of the candidate test scenario;

[0020] Inputting the sequence of candidate test scenario text description words into the Bert model of the test scenario semantic encoder to obtain a sequence of candidate test scenario global text semantic feature vectors;

[0021] Inputting the sequence of candidate test scenario text description words into the LSTM model of the test scenario semantic encoder to obtain a sequence of candidate test scenario context text semantic feature vectors;

[0022] The sequence of the candidate test scenario global text semantic feature vectors and the sequence of the candidate test scenario context text semantic feature vectors are fused to obtain the sequence of the test scenario word granularity semantic encoding feature vectors.

[0023] For example, according to the control method of the cloud-control integrated intelligent driving automatic collaborative test platform of the embodiment of the present application, the sequence of the test scenario word granularity semantic encoding feature vectors is input into the semantic enhancement module using the internal correlation of the sequence to obtain the sequence of the enhanced test scenario word granularity semantic encoding feature vectors, including:

[0024] Determining a first adaptive adjustment factor and a second adaptive adjustment factor;

[0025] Calculating a test scenario word granularity semantic internal autocorrelation matrix of the sequence of the test scenario word granularity semantic encoding feature vectors;

[0026] Performing normalization processing on the test scene word granularity semantic internal autocorrelation matrix based on a soft maximum function to obtain a normalized test scene word granularity semantic internal autocorrelation matrix;

[0027] Perform dot multiplication on the j eigenvalues ​​corresponding to the i-th test scene word granularity semantic coding feature vector in the normalized test scene word granularity semantic internal autocorrelation matrix and the j-th test scene word granularity semantic coding feature vector corresponding thereto, add the sequence of feature vectors obtained after the dot multiplication by position and process them through the sigmoid activation function to obtain the test scene word granularity semantic correlation balanced feature vector, wherein the value of j is from 1 to n, and n is the number of test scene word granularity semantic coding feature vectors in the sequence of the test scene word granularity semantic coding feature vectors;

[0028] Calculate the dot product of the first adaptive adjustment factor and the test scene word granularity semantic relevance balance feature vector and the dot product of the second adaptive adjustment factor and the i-th test scene word granularity semantic coding feature vector, and add them element by element in position to obtain the i-th enhanced test scene word granularity semantic coding feature vector.

[0029] For example, according to the control method of the cloud-control integrated intelligent driving automatic collaborative test platform of an embodiment of the present application, the test scenario word granularity semantic internal autocorrelation matrix of the sequence of the test scenario word granularity semantic encoding feature vectors is calculated, including:

[0030] Cascading the i-th test scene word granularity semantic encoding feature vector and the j-th test scene word granularity semantic encoding feature vector to obtain a test scene word granularity semantic cascade feature vector, wherein i≠j;

[0031] After multiplying the transposed vector of the weight feature vector by the test scenario word granularity semantic cascade feature vector, the obtained value is passed through a nonlinear activation function to obtain the internal autocorrelation between the i-th test scenario word granularity semantic encoding feature vector and the j-th test scenario word granularity semantic encoding feature vector;

[0032] Further, the test scenario word granular semantic internal autocorrelation matrix composed of each of the internal autocorrelations is obtained.

[0033] For example, according to the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform of the embodiment of the present application, the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector is input into the unidirectional attention granularity scanning interactive matching module to obtain a unidirectional matching result representation vector, including:

[0034] Calculating the semantic similarity between the search condition semantic encoding feature vector and each enhanced test scenario word granularity semantic encoding feature vector in the sequence of enhanced test scenario word granularity semantic encoding feature vectors to obtain a sequence of one-way matching semantic measurement coefficients;

[0035] Performing a normalization process based on a soft maximum function on the sequence of one-way matching semantic metric coefficients to obtain a sequence of one-way matching semantic metric weights;

[0036] Using the sequence of the one-way matching semantic measurement weights as weights, calculating the weighted sum of the sequence of the word granularity semantic encoding feature vectors of the enhanced test scenario to obtain a one-way matching interactive reshaping feature vector;

[0037] A differential feature vector between the search condition semantic encoding feature vector and the one-way matching interaction reshaping feature vector is calculated as the one-way matching result representation vector.

[0038] For example, according to the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform of an embodiment of the present application, the semantic similarity between the search condition semantic encoding feature vector and each enhanced test scenario word granularity semantic encoding feature vector in the sequence of the enhanced test scenario word granularity semantic encoding feature vector is calculated to obtain a sequence of one-way matching semantic measurement coefficients, including:

[0039] After cascading the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector, multiplying the weight vector and the cascaded feature vector;

[0040] Add the multiplied value to the bias parameter;

[0041] The added values ​​are subjected to nonlinear activation processing based on a sigmoid function to obtain the one-way matching semantic metric coefficient.

[0042] For example, according to the control method of the cloud-control integrated intelligent driving automatic collaborative test platform of an embodiment of the present application, the search result is determined based on the one-way matching result representation vector, and the search result is used to indicate whether to return the alternative test scenario, including:

[0043] The one-way matching result representation vector is input into a search result generator based on a classifier to obtain the search result.

[0044] For example, the control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to the embodiment of the present application further includes: a training step for training the model;

[0045] Wherein, the training step includes:

[0046] Acquire training data, the training data including: training search conditions, and training text descriptions of candidate test scenarios;

[0047] The training search condition is segmented and then input into the Bert model to obtain a sequence of semantic feature vectors of the training search condition text;

[0048] Cascading the sequence of the training search condition text semantic feature vectors to obtain a training search condition semantic encoding feature vector;

[0049] After word segmentation processing is performed on the training text description of the candidate test scenario, the description is input into a test scenario semantic encoder including a Bert model and an LSTM model to obtain a sequence of word granularity semantic encoding feature vectors of the training test scenario;

[0050] Inputting the sequence of the word granularity semantic encoding feature vectors of the training test scenario into a semantic enhancement module using the internal correlation of the sequence to obtain a sequence of the word granularity semantic encoding feature vectors of the training enhanced test scenario;

[0051] Inputting the sequence of the training search condition semantic encoding feature vector and the training reinforcement test scenario word granularity semantic encoding feature vector into a unidirectional attention granularity scanning interactive matching module to obtain a training unidirectional matching result representation vector;

[0052] Inputting the training one-way matching result representation vector into a search result generator based on a classifier to obtain a classification loss function value;

[0053] The model is trained by back-propagating gradients based on the loss function.

[0054] An embodiment of the present application also provides a cloud-controlled integrated intelligent driving automatic collaborative testing platform, wherein the cloud-controlled integrated intelligent driving automatic collaborative testing platform operates according to any of the control methods described above.

[0055] According to the cloud-controlled integrated intelligent driving automatic collaborative testing platform and control method of the embodiment of the present application, it uses advanced artificial intelligence technology and natural language processing technology to understand the user's test needs and intentions from the search conditions entered by the user, and match them with the text semantic features describing the test scene, so as to return highly matched test scenes to the user. In this way, fully automated matching of massive scenes is achieved to meet the diverse needs of different users for testing the autonomous driving system. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0057] Figure 1 A schematic diagram of the application architecture of a control method for a cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application is shown;

[0058] Figure 2 A flow chart showing a control method of a cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application is shown;

[0059] Figure 3 A flowchart of sub-step S540 of the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application is shown;

[0060] Figure 4 A flowchart of sub-step S550 of the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application is shown;

[0061] Figure 5 A schematic diagram of the structure of a control system of a cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application is shown;

[0062] Figure 6 An application scenario diagram showing a control method of a cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application is shown; and

[0063] Figure 7 A schematic diagram of a storage medium according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0065] The terms used in this specification are those common terms currently widely used in the art in consideration of the functions of the present application, but these terms may vary according to the intention of a person of ordinary skill in the art, precedents, or new technologies in the art. In addition, specific terms may be selected by the applicant, and in this case, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as a general description based on the meaning of the terms and the present application.

[0066] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0067] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0068] Figure 1 A schematic diagram of the application architecture of a control method for a cloud-controlled integrated intelligent driving automatic collaborative testing platform in an embodiment of the present application is shown, including a server 100 and a terminal device 200.

[0069] The terminal device 200 and the server 100 can be connected via the Internet to achieve mutual communication. Optionally, the above-mentioned Internet uses standard communication technology and / or protocol. The Internet is usually the Internet, but it can also be any network, including but not limited to any combination of a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a dedicated network or a virtual private network. In some embodiments, the data exchanged through the network is represented by technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc. In addition, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec) can also be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above-mentioned data communication technologies.

[0070] The server 100 can provide various network services for the terminal device 200, wherein the server 100 can be a single server, a server cluster consisting of several servers, or a cloud computing center. Specifically, the server 100 may include a processor 110 (Center Processing Unit, CPU), a memory 120, an input device 130, and an output device 140, etc. The input device 130 may include a keyboard, a mouse, a touch screen, etc., and the output device 140 may include a display device, such as a liquid crystal display (Liquid Crystal Display, LCD), a cathode ray tube (Cathode Ray Tube, CRT), etc.

[0071] The memory 120 may include a read-only memory (ROM) and a random access memory (RAM), and provides the processor 110 with program instructions and data stored in the memory 120. In an embodiment of the present application, the memory 120 may be used to store a program of a control method of a cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application.

[0072] The processor 110 calls the program instructions stored in the memory 120, and the processor 110 is used to execute the steps of the control method of any cloud-controlled integrated intelligent driving automatic collaborative test platform in the embodiments of the present application according to the obtained program instructions.

[0073] In addition, the application architecture diagram in the embodiment of the present application is intended to more clearly illustrate the technical solution in the embodiment of the present application, and does not constitute a limitation on the technical solution provided in the embodiment of the present application. Of course, for other application architectures and business applications, the technical solution provided in the embodiment of the present application is also applicable to similar problems.

[0074] The following is a non-restrictive explanation of the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform provided according to at least one embodiment of the present application through several examples or embodiments. As described below, different features in these specific examples or embodiments can be combined with each other without conflicting with each other to obtain new examples or embodiments, and these new examples or embodiments also fall within the scope of protection of this application.

[0075] In response to the above technical problems, the technical concept of this application is to build a cloud-controlled integrated intelligent driving automatic collaborative testing platform, which uses advanced artificial intelligence technology and natural language processing technology to understand the user's test needs and intentions from the search conditions entered by the user, and match them with the text semantic features describing the test scenario, so as to return highly matched test scenarios to the user. In this way, fully automated matching of massive scenarios is achieved to meet the diverse needs of different users for autonomous driving system testing.

[0076] Figure 2 The flowchart of the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform in the embodiment of the present application is shown. For example, the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform can be executed by a server, which can be Figure 1 The server 100 shown in FIG. Figure 2As shown, the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform according to the embodiment of the present application includes the steps of: S510, obtaining the search conditions input by the user; S520, extracting the text description of the alternative test scenario from the test scenario database; S530, semantically encoding the search conditions to obtain the search condition semantic encoding feature vector; S540, semantically encoding and internally related semantically enhancing the text description of the alternative test scenario to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors; S550, inputting the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector into the unidirectional attention granularity scanning interactive matching module to obtain a unidirectional matching result representation vector; S560, determining the search result based on the unidirectional matching result representation vector, and the search result is used to indicate whether to return the alternative test scenario.

[0077] Based on this, in the technical solution of the present application, first, the search conditions input by the user are obtained; and the text description of the alternative test scenarios is extracted from the test scenario database. Among them, the search conditions are obtained in order to clarify the user's test needs. Specifically, the search conditions usually include the user's specific needs and expectations for the test scenarios, such as specific traffic environments, driving behaviors or functional tests. That is, obtaining the search conditions helps the platform to understand the user's test purposes more accurately. In addition, the test scenario database stores a large number of different test scenarios and text descriptions corresponding to each test scenario, and these text descriptions usually contain specific information about the test scenarios, such as traffic environments, driving conditions, and specific events that may occur. Extracting the text description of the alternative test scenarios in the technical solution of the present application is an important prerequisite for the subsequent understanding of the test environment information of the alternative test scenarios.

[0078] Then, the search condition is semantically encoded to obtain a search condition semantic encoding feature vector, that is, the semantic features of the search condition are captured by semantic encoding to deeply understand the user's test intention and test requirements contained in the search condition.

[0079] Correspondingly, in step S530, the search condition is semantically encoded to obtain a search condition semantic encoding feature vector, including: segmenting the search condition and inputting it into the Bert model to obtain a sequence of search condition text semantic feature vectors; and cascading the sequence of search condition text semantic feature vectors to obtain the search condition semantic encoding feature vector.

[0080] At the same time, considering that the text description of the alternative test scenario may have a complex semantic feature representation, in the technical solution of the present application, the text description of the alternative test scenario is processed by word segmentation and then input into the test scenario semantic encoder containing the Bert model and the LSTM model to obtain a sequence of test scenario word granularity semantic encoding feature vectors. In particular, the Bert model is mainly composed of the encoding layer of the Transformer, and its encoding layer is mainly based on the self-attention structure. This structure enables the Bert model to capture the global information of the text sequence while parallelizing the training, but it also makes it difficult to obtain the order and grammatical relationship between the words in the text sequence. Although the position and order of words can be marked by position embedding, it still has no advantage over the traditional LSTM network structure in obtaining the contextual representation of words. Therefore, in the technical solution of the present application, it is considered to combine the Bert model and the LSTM model to capture the global semantic features of the text of the alternative test scenario while capturing its contextual semantic dependency association relationship, so as to make full use of the advantages of both.

[0081] Next, the sequence of the test scene word granularity semantic coding feature vectors is input into a semantic enhancement module that utilizes the internal correlation of the sequence to further strengthen the semantic coherence association and syntactic dependency association relationship between the test scene words, thereby obtaining a sequence of enhanced test scene word granularity semantic coding feature vectors. In an embodiment of the present application, the specific process of inputting the sequence of the test scene word granularity semantic coding feature vectors into a semantic enhancement module that utilizes the internal correlation of the sequence to obtain a sequence of enhanced test scene word granularity semantic coding feature vectors includes: first determining a first adaptive adjustment factor and a second adaptive adjustment factor. Among them, the introduction of the first adaptive adjustment factor and the second adaptive adjustment factor can make the semantic enhancement process adaptive, so that the intensity of semantic enhancement can be dynamically adjusted. Next, the test scene word granularity semantic internal autocorrelation matrix of the sequence of the test scene word granularity semantic coding feature vectors is calculated to characterize the correlation between each test scene word granularity semantic coding feature vector and other test scene word granularity semantic coding feature vectors, which is the key to understanding the internal semantic correlation of the sequence. Then, the test scene word granularity semantic internal autocorrelation matrix is ​​normalized based on the soft maximum function to obtain the normalized test scene word granularity semantic internal autocorrelation matrix; and the j eigenvalues ​​corresponding to the i-th test scene word granularity semantic coding feature vector in the normalized test scene word granularity semantic internal autocorrelation matrix are respectively multiplied with the j-th test scene word granularity semantic coding feature vector corresponding to it, and the sequence of feature vectors obtained after the dot multiplication is added by position and processed by the sigmoid activation function to obtain the test scene word granularity semantic correlation balanced feature vector, wherein the value of j is one to n, and n is the number of test scene word granularity semantic coding feature vectors in the sequence of the test scene word granularity semantic coding feature vectors. It should be understood that in this way, the semantic feature distribution with a high correlation with the i-th test scene word granularity semantic coding feature vector can be highlighted, which helps to concentrate and strengthen the features with a high correlation in the sequence. After that, the dot product of the first adaptive adjustment factor and the test scene word granularity semantic correlation equilibrium feature vector and the dot product of the second adaptive adjustment factor and the i-th test scene word granularity semantic coding feature vector are calculated and added element by element to obtain the i-th enhanced test scene word granularity semantic coding feature vector. In this way, the sequence of the enhanced test scene word granularity semantic coding feature vectors composed of the individual enhanced test scene word granularity semantic coding feature vectors can better reflect the semantic structure and overall correlation within the sequence.

[0082] Accordingly, in step S540, if Figure 3As shown, the text description of the alternative test scenario is semantically encoded and internally correlated semantically enhanced to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors, including: S541, after word segmentation processing, the text description of the alternative test scenario is input into a test scenario semantic encoder including a Bert model and an LSTM model to obtain a sequence of test scenario word granularity semantic encoding feature vectors; S542, the sequence of test scenario word granularity semantic encoding feature vectors is input into a semantic enhancement module using the internal correlation of the sequence to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors.

[0083] Wherein, in step S541, the text description of the alternative test scenario is segmented and then input into the test scenario semantic encoder including the Bert model and the LSTM model to obtain a sequence of test scenario word granularity semantic coding feature vectors, including: segmenting the text description of the alternative test scenario to obtain a sequence of alternative test scenario text description words; inputting the sequence of alternative test scenario text description words into the Bert model of the test scenario semantic encoder to obtain a sequence of alternative test scenario global text semantic feature vectors; inputting the sequence of alternative test scenario text description words into the LSTM model of the test scenario semantic encoder to obtain a sequence of alternative test scenario context text semantic feature vectors; fusing the sequence of alternative test scenario global text semantic feature vectors and the sequence of alternative test scenario context text semantic feature vectors to obtain a sequence of test scenario word granularity semantic coding feature vectors. It should be understood that the fusion can be performed by cascading or adding by position.

[0084] Among them, in step S542, the sequence of the test scenario word granularity semantic coding feature vectors is input into the semantic enhancement module using the internal correlation of the sequence to obtain the sequence of the enhanced test scenario word granularity semantic coding feature vectors, including: determining a first adaptive adjustment factor and a second adaptive adjustment factor; calculating the test scenario word granularity semantic internal autocorrelation matrix of the sequence of the test scenario word granularity semantic coding feature vectors; normalizing the test scenario word granularity semantic internal autocorrelation matrix based on the soft maximum function to obtain the normalized test scenario word granularity semantic internal autocorrelation matrix; and adding j features corresponding to the i-th test scenario word granularity semantic coding feature vector in the normalized test scenario word granularity semantic internal autocorrelation matrix. The eigenvalues ​​are respectively dot-multiplied with the corresponding j-th test scenario word granularity semantic coding feature vectors, and the sequence of feature vectors obtained after the dot multiplication is added positionally and processed through the sigmoid activation function to obtain a test scenario word granularity semantic correlation balanced feature vector, wherein the value of j is from one to n, and n is the number of test scenario word granularity semantic coding feature vectors in the sequence of test scenario word granularity semantic coding feature vectors; the dot product of the first adaptive adjustment factor and the test scenario word granularity semantic correlation balanced feature vector and the dot product of the second adaptive adjustment factor and the i-th test scenario word granularity semantic coding feature vector are calculated and added element by element positionally to obtain the i-th enhanced test scenario word granularity semantic coding feature vector.

[0085] Specifically, the test scenario word granularity semantic internal autocorrelation matrix of the sequence of the test scenario word granularity semantic encoding feature vectors is calculated, including: cascading the i-th test scenario word granularity semantic encoding feature vector and the j-th test scenario word granularity semantic encoding feature vector to obtain a test scenario word granularity semantic cascade feature vector, where i≠j; multiplying the transpose vector of the weight feature vector by the test scenario word granularity semantic cascade feature vector, and then passing the obtained value through a nonlinear activation function to obtain the internal autocorrelation between the i-th test scenario word granularity semantic encoding feature vector and the j-th test scenario word granularity semantic encoding feature vector; and further obtaining the test scenario word granularity semantic internal autocorrelation matrix composed of each of the internal autocorrelations.

[0086] In a specific example, the sequence of the test scenario word granularity semantic encoding feature vectors is input into a semantic enhancement module using sequence internal correlation to obtain the sequence of the enhanced test scenario word granularity semantic encoding feature vectors, including: the sequence of the test scenario word granularity semantic encoding feature vectors is input into a semantic enhancement module using sequence internal correlation to process the sequence with the following enhancement formula to obtain the sequence of the enhanced test scenario word granularity semantic encoding feature vectors; wherein the enhancement formula is:

[0087] Sij =f1(V r T ·[v i ||v j(j≠i) ])

[0088]

[0089] Among them, v i and v j are the i-th and j-th test scene word granularity semantic encoding feature vectors in the sequence of the test scene word granularity semantic encoding feature vectors, respectively. [·‖·] represents the cascade of vectors. V r T is the transposed vector of the weight feature vector, f1 represents the nonlinear activation function, S ij is the eigenvalue at position (i, j) in the semantic internal autocorrelation matrix of the test scene word granularity, exp(·) represents an exponential function with a natural constant as the base, λ is the inverse temperature coefficient of the softmax function, α ij is the eigenvalue at position (i, j) in the normalized test scene word granularity semantic internal autocorrelation matrix, n is the number of test scene word granularity semantic encoding feature vectors in the sequence of test scene word granularity semantic encoding feature vectors, σ represents the sigmoid activation function, α and β are the first adaptive adjustment factor and the second adaptive adjustment factor, * represents the dot product, v′ i is the ith enhanced test scenario word granularity semantic encoding feature vector in the sequence of the enhanced test scenario word granularity semantic encoding feature vectors. Further, f1 represents a nonlinear activation function, such as a Sigmoid function and a Tanh function.

[0090] Subsequently, the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector is input into the one-way attention granularity scanning interactive matching module to obtain a one-way matching result representation vector. The one-way attention granularity scanning interactive matching module quantifies the semantic similarity between the search condition semantic encoding feature vector and each enhanced test scenario word granularity semantic encoding feature vector as a fine-grained one-way matching semantic measurement coefficient between the search condition and the text description of the alternative test scenario, so as to use the search condition semantic encoding feature vector as a search word to perform granularity scanning interactive search matching on the text description of the alternative test scenario, thereby obtaining a one-way matching result representation vector.

[0091] Accordingly, in step S550, if Figure 4As shown, the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector is input into the one-way attention granularity scanning interactive matching module to obtain a one-way matching result representation vector, including: S551, calculating the semantic similarity between the search condition semantic encoding feature vector and each enhanced test scenario word granularity semantic encoding feature vector in the sequence of the enhanced test scenario word granularity semantic encoding feature vector to obtain a sequence of one-way matching semantic measurement coefficients; S552, performing a normalization process based on a soft maximum function on the sequence of one-way matching semantic measurement coefficients to obtain a sequence of one-way matching semantic measurement weights; S553, using the sequence of one-way matching semantic measurement weights as weights, calculating the weighted sum of the sequence of the enhanced test scenario word granularity semantic encoding feature vectors to obtain a one-way matching interactive reshaping feature vector; S554, calculating the differential feature vector between the search condition semantic encoding feature vector and the one-way matching interactive reshaping feature vector as the one-way matching result representation vector.

[0092] Among them, in step S551, the semantic similarity between the search condition semantic coding feature vector and each enhanced test scenario word granularity semantic coding feature vector in the sequence of the enhanced test scenario word granularity semantic coding feature vector is calculated to obtain a sequence of one-way matching semantic measurement coefficients, including: after cascading the search condition semantic coding feature vector and the enhanced test scenario word granularity semantic coding feature vector, multiplying the weight vector and the cascaded feature vector; adding the multiplied value and the bias parameter; and performing nonlinear activation processing based on the sigmoid function on the added value to obtain the one-way matching semantic measurement coefficient.

[0093] In a specific example, the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector is input into a one-way attention granularity scanning interactive matching module to obtain a one-way matching result representation vector, including: inputting the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector into the one-way attention granularity scanning interactive matching module and processing them with the following matching formula to obtain the one-way matching result representation vector; wherein the matching formula is:

[0094] r 1j =σ(W[v1,v j ]+b)

[0095]

[0096] Where v1 is the semantic encoding feature vector of the search condition, v jis the jth enhanced test scenario word granularity semantic encoding feature vector of the sequence of the enhanced test scenario word granularity semantic encoding feature vectors, W and b are the weight vector and bias parameter respectively, σ represents the sigmoid function, r 1j is the j-th one-way matching semantic metric coefficient, exp(·) represents an exponential function with a natural constant as the base, the value of m is the number of enhanced test scenario word granularity semantic encoding feature vectors in the sequence of the enhanced test scenario word granularity semantic encoding feature vectors, e 1j is the j-th one-way matching semantic metric weight, v 1p is the one-way matching interaction reshape feature vector, v m is the one-way matching result representation vector.

[0097] Furthermore, the one-way matching result representation vector is input into a search result generator based on a classifier to obtain a search result, and the search result is used to indicate whether to return the alternative test scenario.

[0098] Accordingly, in step S560, a search result is determined based on the one-way matching result representation vector, and the search result is used to indicate whether to return the alternative test scenario, including: inputting the one-way matching result representation vector into a classifier-based search result generator to obtain the search result.

[0099] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and classifiers, and then classify (or predict) unknown data. Logistic regression and SVM are often used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are required to form a multi-classification. However, this is prone to errors and is not efficient. Commonly used multi-classification methods include the Softmax classification function.

[0100] In a specific example, the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform also includes: a training step for training the model; wherein the training step includes: obtaining training data, the training data including: training search conditions, and training text descriptions of alternative test scenarios; segmenting the training search conditions and inputting them into the Bert model to obtain a sequence of training search condition text semantic feature vectors; cascading the sequence of training search condition text semantic feature vectors to obtain training search condition semantic encoding feature vectors; segmenting the training text descriptions of the alternative test scenarios and inputting them into the test scene containing the Bert model and the LSTM model; The method comprises the following steps: inputting the sequence of word-granularity semantic encoding feature vectors of the training test scene into a semantic enhancement module that utilizes the internal correlation of the sequence to obtain a sequence of word-granularity semantic encoding feature vectors of the training enhanced test scene; inputting the sequence of the training search condition semantic encoding feature vectors and the training enhanced test scene word-granularity semantic encoding feature vectors into a unidirectional attention granularity scanning interactive matching module to obtain a training unidirectional matching result representation vector; inputting the training unidirectional matching result representation vector into a search result generator based on a classifier to obtain a classification loss function value; and training the model through gradient back propagation based on the loss function.

[0101] In the technical solution of the present application, the training search condition semantic coding feature vector represents the text semantic features of the training search condition, and the sequence of the training intensive test scenario word granularity semantic coding feature vectors represents the semantic feature distribution of the intensive word granularity text semantic context association features of the training text description of the alternative test scenario. When the training search condition semantic coding feature vector and the sequence of the training intensive test scenario word granularity semantic coding feature vector are input into the one-way attention granularity scanning interactive matching module, the semantic association between the training search condition semantic coding feature vector and each training intensive test scenario word granularity semantic coding feature vector in the sequence of the training intensive test scenario word granularity semantic coding feature vector is used as the local semantic matching factor, and the local semantic matching factor is used as the class attention weight to comprehensively represent the one-way attention granularity scanning interactive matching result of the training search condition semantic coding feature vector relative to the sequence of the training intensive test scenario word granularity semantic coding feature vector, that is, the training one-way matching result representation vector. However, considering that the feature sparsity of each training reinforcement test scene word granularity semantic coding feature vector in the sequence of the training search condition semantic coding feature vector and the training reinforcement test scene word granularity semantic coding feature vector is different, this will lead to the technical problem that the training one-way matching result representation vector obtained by interactive matching has significant local feature distribution heterogeneity as a whole feature distribution, which leads to the class regression recognition problem of one-way attention granularity scanning interactive matching, thereby affecting the classification training efficiency.

[0102] Therefore, in the model training process of the present application, a predetermined loss function other than the classification loss function value is further introduced, that is, the model is trained by gradient back propagation based on the loss function, including the following steps:

[0103] Calculate a first one-way matching result weight matrix and a second one-way matching result weight matrix based on the training one-way matching result representation vector, wherein the eigenvalues ​​of the (i, j)th positions of the first one-way matching result weight matrix and the second one-way matching result weight matrix are respectively half of the mean and the absolute value of the difference of the i-th eigenvalue and the j-th eigenvalue of the training one-way matching result representation vector;

[0104] Performing query matrix multiplication of the training one-way matching result representation vector with the first one-way matching result weight matrix and the second one-way matching result weight matrix respectively to obtain a first one-way matching result intermediate vector and a second one-way matching result intermediate vector;

[0105] Calculate the vector inner product of the first one-way matching result intermediate vector and the second one-way matching result intermediate vector to obtain a first one-way matching result loss term;

[0106] Performing matrix multiplication of the first one-way matching result weight matrix and the second one-way matching result weight matrix, and calculating the Frobenius norm of the result matrix to obtain a second one-way matching result loss term;

[0107] Subtract the product of a predetermined weight hyperparameter and the second one-way matching result loss term from the first one-way matching result loss term to obtain a one-way matching result loss function value;

[0108] Based on the weighted sum of the one-way matching result loss function value and the classification loss function value, the model parameters are optimized through gradient back propagation.

[0109] The one-way matching result loss function value is specifically expressed as:

[0110]

[0111]

[0112] Wherein, V represents the training one-way matching result representation vector, v i and v j Respectively represent the i-th eigenvalue and j-th eigenvalue of the training one-way matching result representation vector, M μ Represents the weight matrix of the first one-way matching result, M σ Represents the second one-way matching result weight matrix, M μ (i, j) represents the eigenvalue of the (i, j)th position of the first one-way matching result weight matrix, M σ (i,j) represents the eigenvalue of the (i,j)th position of the second one-way matching result weight matrix, represents matrix multiplication, T represents transpose, α represents a predetermined weight hyperparameter, ‖·‖ F represents the norm of the matrix, and Loss represents the loss function value of the one-way matching result.

[0113] That is, the one-way matching result loss function value uses the structured feature representation of the short-range and long-range cross-scale detail links of the training one-way matching result representation vector to perform the query composition of the detail inner product space within the training one-way matching result representation vector, so as to approximate the low-rank independent observable composition of the link detail composition provided by the structured detail interaction of the training one-way matching result representation vector. In this way, by training with the training one-way matching result loss function value, the distributed detail groups of the training one-way matching result representation vector can be used to decompose the detail groups based on the complexity of the details, so as to promote the class regression decomposition and recognition of the complex feature structure of the training one-way matching result representation vector and improve the classification training efficiency.

[0114] Furthermore, the present application also provides a cloud-controlled integrated intelligent driving automatic collaborative testing platform, which operates according to any of the control methods described above.

[0115] Based on the above embodiments, see Figure 5 As shown, it is a structural schematic diagram of a control system 800 of a cloud-controlled integrated intelligent driving automatic collaborative test platform in an embodiment of the present application. The control system 800 of the cloud-controlled integrated intelligent driving automatic collaborative test platform includes: a search condition acquisition module 810, which is used to obtain the search condition input by the user; a test data extraction module 820, which is used to extract the text description of the alternative test scenario from the test scenario database; a semantic encoding module 830, which is used to semantically encode the search condition to obtain a search condition semantic encoding feature vector; an enhancement module 840, which is used to semantically encode the text description of the alternative test scenario and enhance the internal related semantics to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors; a one-way matching module 850, which is used to input the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector into a one-way attention granularity scanning interactive matching module to obtain a one-way matching result representation vector; a search result determination module 860, which is used to determine the search result based on the one-way matching result representation vector, and the search result is used to indicate whether to return the alternative test scenario.

[0116] Here, those skilled in the art can understand that the specific functions and operations of each module in the control system 800 of the above-mentioned cloud-controlled integrated intelligent driving automatic collaborative test platform have been referred to above. Figures 2 to 4 The control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform has been introduced in detail, and therefore, its repeated description will be omitted.

[0117] Figure 6 FIG. 1 is an application scenario diagram of a control method for a cloud-controlled integrated intelligent driving automatic collaborative test platform according to an embodiment of the present application. Figure 6 As shown, in this application scenario, first, the search condition entered by the user is obtained (for example, Figure 6 D1 shown in FIG) and extracting text descriptions of candidate test scenarios from a test scenario database (e.g., Figure 6 Then, the search condition and the text description of the candidate test scenario are input into a server (for example, Figure 6In S) as shown in , the server can use the control algorithm of the cloud-controlled integrated intelligent driving automatic collaborative testing platform to process the search conditions and the text description of the alternative test scenario to obtain a search result indicating whether to return the alternative test scenario.

[0118] Based on the above embodiments, another exemplary embodiment of an electronic device is also provided in the embodiments of the present application. In some possible implementations, the electronic device in the embodiments of the present application may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform in the above embodiments can be implemented when the processor executes the program.

[0119] For example, in the case of electronic equipment Figure 1 Taking the server 100 in the example as an example, the processor in the electronic device is the processor 110 in the server 100, and the memory in the electronic device is the memory 120 in the server 100.

[0120] An embodiment of the present application also provides a computer-readable storage medium. Figure 7 1 shows a schematic diagram of a computer-readable storage medium 1000 according to an embodiment of the present application. Figure 7 As shown, the computer-readable storage medium 1000 stores computer-executable instructions 1001. When the computer-executable instructions 1001 are executed by the processor, the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform according to the embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.

[0121] The embodiment of the present application also provides a computer program product or computer program, which includes computer executable instructions, and the computer executable instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer executable instructions from the computer-readable storage medium, and the processor executes the computer executable instructions, so that the computer device executes the control method of the cloud-controlled integrated intelligent driving automatic collaborative test platform according to the embodiment of the present application.

[0122] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various variations and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.

[0123] In addition, although the present application makes various references to certain units in the system according to embodiments of the present application, any number of different units can be used and run on the client and / or server. The units are only illustrative, and different aspects of the system and method can use different units.

[0124] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present application is not limited to any particular form of combination of hardware and software.

[0125] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0126] The above is an explanation of the present application and should not be considered as a limitation thereof. Although several exemplary embodiments of the present application have been described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all of these modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the above is an explanation of the present application and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. A control method for a cloud-controlled integrated intelligent driving automatic collaborative test platform, characterized in that: include: Get the search criteria entered by the user; Extracting text descriptions of candidate test scenarios from a test scenario database; Performing semantic encoding on the search condition to obtain a search condition semantic encoding feature vector; Performing semantic encoding and internal related semantic enhancement on the text description of the candidate test scenario to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors; Inputting the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector into a unidirectional attention granularity scanning interactive matching module to obtain a unidirectional matching result representation vector; Determine a search result based on the one-way matching result representation vector, wherein the search result is used to indicate whether to return the alternative test scenario; The method further comprises: a training step for training the model; Wherein, the training step includes: Acquire training data, the training data including: training search conditions, and training text descriptions of candidate test scenarios; The training search condition is segmented and then input into the Bert model to obtain a sequence of semantic feature vectors of the training search condition text; Cascading the sequence of the training search condition text semantic feature vectors to obtain a training search condition semantic encoding feature vector; After word segmentation processing is performed on the training text description of the candidate test scenario, the description is input into a test scenario semantic encoder including a Bert model and an LSTM model to obtain a sequence of word granularity semantic encoding feature vectors of the training test scenario; Inputting the sequence of the word granularity semantic encoding feature vectors of the training test scenario into a semantic enhancement module using the internal correlation of the sequence to obtain a sequence of the word granularity semantic encoding feature vectors of the training enhanced test scenario; Inputting the sequence of the training search condition semantic encoding feature vector and the training reinforcement test scenario word granularity semantic encoding feature vector into a unidirectional attention granularity scanning interactive matching module to obtain a training unidirectional matching result representation vector; Inputting the training one-way matching result representation vector into a search result generator based on a classifier to obtain a classification loss function value; Train the model through gradient backpropagation based on the loss function; The method of training the model by gradient back propagation based on the loss function includes the following steps: Calculate a first one-way matching result weight matrix and a second one-way matching result weight matrix based on the training one-way matching result representation vector, wherein the eigenvalues ​​of the (i, j)th positions of the first one-way matching result weight matrix and the second one-way matching result weight matrix are respectively half of the mean and the absolute value of the difference of the i-th eigenvalue and the j-th eigenvalue of the training one-way matching result representation vector; Performing query matrix multiplication of the training one-way matching result representation vector with the first one-way matching result weight matrix and the second one-way matching result weight matrix respectively to obtain a first one-way matching result intermediate vector and a second one-way matching result intermediate vector; Calculate the vector inner product of the first one-way matching result intermediate vector and the second one-way matching result intermediate vector to obtain a first one-way matching result loss term; Performing matrix multiplication of the first one-way matching result weight matrix and the second one-way matching result weight matrix, and calculating the Frobenius norm of the result matrix to obtain a second one-way matching result loss term; Subtract the product of a predetermined weight hyperparameter and the second one-way matching result loss term from the first one-way matching result loss term to obtain a one-way matching result loss function value; Based on the weighted sum of the one-way matching result loss function value and the classification loss function value, the model parameters are optimized through gradient back propagation.

2. The control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to claim 1 is characterized in that: Semantically encoding the search condition to obtain a search condition semantic encoding feature vector includes: The search condition is segmented and then input into the Bert model to obtain a sequence of semantic feature vectors of the search condition text; The sequence of the search condition text semantic feature vectors is cascaded to obtain the search condition semantic encoding feature vector.

3. The control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to claim 2 is characterized in that: The text description of the candidate test scenario is semantically encoded and internally related semantically enhanced to obtain a sequence of enhanced test scenario word granularity semantic encoding feature vectors, including: After word segmentation processing is performed on the text description of the candidate test scenario, the text description is input into a test scenario semantic encoder including a Bert model and an LSTM model to obtain a sequence of test scenario word granularity semantic encoding feature vectors; The sequence of the test scenario word granularity semantic encoding feature vectors is input into a semantic enhancement module that uses the internal correlation of the sequence to obtain the sequence of enhanced test scenario word granularity semantic encoding feature vectors.

4. The control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to claim 3 is characterized in that: After word segmentation processing is performed on the text description of the candidate test scenario, the text description is input into the test scenario semantic encoder including the Bert model and the LSTM model to obtain a sequence of test scenario word granularity semantic encoding feature vectors, including: Segmenting the text description of the candidate test scenario to obtain a sequence of text description words of the candidate test scenario; Inputting the sequence of candidate test scenario text description words into the Bert model of the test scenario semantic encoder to obtain a sequence of candidate test scenario global text semantic feature vectors; Inputting the sequence of candidate test scenario text description words into the LSTM model of the test scenario semantic encoder to obtain a sequence of candidate test scenario context text semantic feature vectors; The sequence of the candidate test scenario global text semantic feature vectors and the sequence of the candidate test scenario context text semantic feature vectors are fused to obtain the sequence of the test scenario word granularity semantic encoding feature vectors.

5. The control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to claim 4 is characterized in that: Inputting the sequence of the test scenario word granularity semantic encoding feature vectors into a semantic enhancement module using the internal correlation of the sequence to obtain the sequence of the enhanced test scenario word granularity semantic encoding feature vectors, including: Determining a first adaptive adjustment factor and a second adaptive adjustment factor; Calculating a test scenario word granularity semantic internal autocorrelation matrix of the sequence of the test scenario word granularity semantic encoding feature vectors; Performing normalization processing on the test scene word granularity semantic internal autocorrelation matrix based on a soft maximum function to obtain a normalized test scene word granularity semantic internal autocorrelation matrix; Perform dot multiplication on the j eigenvalues ​​corresponding to the i-th test scene word granularity semantic coding feature vector in the normalized test scene word granularity semantic internal autocorrelation matrix and the j-th test scene word granularity semantic coding feature vector corresponding thereto, add the sequence of feature vectors obtained after the dot multiplication by position and process them through the sigmoid activation function to obtain the test scene word granularity semantic correlation balanced feature vector, wherein the value of j is from 1 to n, and n is the number of test scene word granularity semantic coding feature vectors in the sequence of the test scene word granularity semantic coding feature vectors; Calculate the dot product of the first adaptive adjustment factor and the test scene word granularity semantic relevance balance feature vector and the dot product of the second adaptive adjustment factor and the i-th test scene word granularity semantic coding feature vector, and add them element by element in position to obtain the i-th enhanced test scene word granularity semantic coding feature vector.

6. The control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to claim 5 is characterized in that: Calculating the test scenario word granularity semantic internal autocorrelation matrix of the sequence of the test scenario word granularity semantic encoding feature vectors, including: Cascading the i-th test scene word granularity semantic encoding feature vector and the j-th test scene word granularity semantic encoding feature vector to obtain a test scene word granularity semantic cascade feature vector, wherein i≠j; After multiplying the transposed vector of the weight feature vector by the test scenario word granularity semantic cascade feature vector, the obtained value is passed through a nonlinear activation function to obtain the internal autocorrelation between the i-th test scenario word granularity semantic encoding feature vector and the j-th test scenario word granularity semantic encoding feature vector; Further, the test scenario word granular semantic internal autocorrelation matrix composed of each of the internal autocorrelations is obtained.

7. The control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to claim 6 is characterized in that: Inputting the sequence of the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector into a unidirectional attention granularity scanning interactive matching module to obtain a unidirectional matching result representation vector, including: Calculating the semantic similarity between the search condition semantic encoding feature vector and each enhanced test scenario word granularity semantic encoding feature vector in the sequence of enhanced test scenario word granularity semantic encoding feature vectors to obtain a sequence of one-way matching semantic measurement coefficients; Performing a normalization process based on a soft maximum function on the sequence of one-way matching semantic metric coefficients to obtain a sequence of one-way matching semantic metric weights; Using the sequence of the one-way matching semantic measurement weights as weights, calculating the weighted sum of the sequence of the word granularity semantic encoding feature vectors of the enhanced test scenario to obtain a one-way matching interactive reshaping feature vector; Calculating a differential feature vector between the search condition semantic encoding feature vector and the one-way matching interactive reshaping feature vector as the one-way matching result representation vector; The step of calculating the semantic similarity between the search condition semantic encoding feature vector and each enhanced test scenario word granularity semantic encoding feature vector in the sequence of enhanced test scenario word granularity semantic encoding feature vectors to obtain a sequence of one-way matching semantic measurement coefficients includes: After cascading the search condition semantic encoding feature vector and the enhanced test scenario word granularity semantic encoding feature vector, multiplying the weight vector and the cascaded feature vector; Add the multiplied value to the bias parameter; The added values ​​are subjected to nonlinear activation processing based on a sigmoid function to obtain the one-way matching semantic metric coefficient.

8. The control method of the cloud-control integrated intelligent driving automatic collaborative test platform according to claim 7 is characterized in that: Determining a search result based on the one-way matching result representation vector, where the search result is used to indicate whether to return the alternative test scenario, includes: The one-way matching result representation vector is input into a search result generator based on a classifier to obtain the search result.

9. A cloud-controlled integrated intelligent driving automatic collaborative testing platform, characterized in that: The cloud-controlled integrated intelligent driving automatic collaborative testing platform operates according to the control method described in any one of claims 1 to 8.

Images