Simulation service platform data generation optimization method, system, equipment and medium

Abstract

The invention discloses a simulation service platform data generation optimization method, system and device and a medium, belongs to the field of data generation optimization processing, and solves the problems that in an existing data generation mode, models are single, and complex relations are difficult to simulate. The method comprises the following steps: an encoder quantizes, encodes and compresses a first input data vector to obtain a first compression distance; the discrimination neural network carries out category discrimination on the first compression distance; the generation neural network generates associated data of the first compression distance; the decoder decompresses and decodes the associated data to obtain a second input data vector; the encoder quantizes and encodes the second input data vector, and calculates a compression distance between the first input data vector and the second input data vector to obtain a third compression distance; the discrimination neural network carries out category discrimination on the third compression distance; and taking the second input data vector as a data generation result of the first input data vector if the category judgment result of the third compression distance and the first compression distance is one category.

Description

Technical Field

[0001] The present invention relates to the technical field of data generation optimization processing, and in particular to a simulation service platform data generation optimization method, system, device and medium. Background Art

[0002] With the rapid development of information technology and industrial design, simulation service platforms are playing an increasingly important role in product development, simulation testing, and optimized design. These platforms face many challenges in data management and processing, which has promoted the continuous innovation and development of related technologies.

[0003] Data generation is an important part of the data processing process. Traditional data generation methods rely on a single neural network model, or statistical methods such as random sampling, statistical simulation, or deterministic algorithms. This method does not deeply explore the characteristics of related data, is difficult to simulate complex relationships, and has poor flexibility, resulting in less than ideal data generation results. Summary of the invention

[0004] In view of the above analysis, the embodiments of the present invention aim to provide a simulation service platform data generation optimization method, system, device and medium to solve the problems of existing data generation methods such as single model and difficulty in simulating complex relationships.

[0005] The present invention discloses a simulation service platform data generation optimization method, the method comprising: The encoder quantizes, encodes and compresses the first input data vector to obtain a first compression distance; The discriminant neural network performs category discrimination on the first compression distance to obtain a category discrimination result of the first compression distance; The generating neural network generates associated data of the first compression distance based on the first compression distance and the category discrimination result thereof; the decoder decompresses and decodes the associated data of the first compression distance to obtain a second input data vector; The encoder quantizes and encodes the second input data vector, and calculates a compression distance between the quantized and encoded first input data vector and the second input data vector to obtain a third compression distance; The discriminant neural network performs category discrimination on the third compression distance to obtain a category discrimination result of the third compression distance; if the category discrimination results of the third compression distance and the first compression distance are the same, the second input data vector is used as the data generation result of the first input data vector; otherwise, jump to the step of generating the associated data of the first compression distance based on the first compression distance and its category discrimination result by the generating neural network, and regenerate the second input data vector.

[0006] On the basis of the above scheme, the present invention also makes the following improvements: Further, the calculation of the compression distance between the first input data vector and the second input data vector after quantization encoding is performed: Separately compressing the first input data vector and the second input data vector after quantization encoding, and recording the sizes of the compressed files of the respective compressed files; Combining the first input data vector and the second input data vector after quantization coding, and recording the size of the compressed file of the combined compression; The compression distance between the quantized encoded first input data vector and the second input data vector is calculated according to the sizes of the compressed files of the first input data vector and the second input data vector compressed individually and the size of the compressed file compressed jointly.

[0007] Further, the compression distance CD between the first input data vector and the second input data vector after quantization coding is expressed as: , in, , Respectively represent the sizes of compressed files of the first input data vector and the second input data vector after quantization encoding; Indicates the size of the compressed file obtained by jointly compressing the first input data vector and the second input data vector after quantization encoding.

[0008] Furthermore, before jointly compressing the first input data vector and the second input data vector after quantization encoding, the following steps are further performed: Concatenate the quantized and encoded first input data vector and the second input data vector into a long string; Then, the quantized and encoded first input data vector and the second input data vector are jointly compressed.

[0009] Furthermore, before the encoder quantizes the first input data vector and the second input data vector, the following steps are performed: Preprocessing operations are performed on the first input data vector and the second input data vector respectively.

[0010] Furthermore, the preprocessing operation includes normalization, denoising and enhancement.

[0011] Further, the encoder quantizes and encodes the first input data vector and the second input data vector, and performs: The encoder multiplies each element in the first input data vector and the second input data vector by a quantization factor, and rounds to the nearest integer. Then, the encoding is converted into a binary string of fixed length to obtain the first input data vector and the second input data vector after quantization encoding.

[0012] The present invention also provides a simulation service platform data generation optimization system, the system comprising an encoder, a discriminant neural network, a generative neural network and a decoder; wherein: The encoder is used to quantize, encode and compress the first input data vector to obtain a first compression distance; and is also used to quantize and encode the second input data vector output by the decoder, and calculate the compression distance between the quantized and encoded first input data vector and the second input data vector to obtain a third compression distance; A discriminant neural network is used to perform category discrimination on the first compression distance to obtain a category discrimination result of the first compression distance; and is also used to perform category discrimination on the third compression distance to obtain a category discrimination result of the third compression distance; if the category discrimination results of the third compression distance and the first compression distance are the same, the second input data vector is used as the data generation result of the first input data vector; otherwise, the generation neural network is instructed to regenerate the associated data of the first compression distance based on the first compression distance and its category discrimination result, until the category discrimination results of the third compression distance and the first compression distance are the same; Generate a neural network for generating associated data of the first compression distance based on the first compression distance and the category discrimination result thereof; The decoder is used to decompress and decode the associated data of the first compression distance to obtain a second input data vector.

[0013] The present invention also provides an electronic device, comprising: Processor; and A memory stores executable codes thereon, and when the executable codes are executed by the processor, the processor executes the above-mentioned simulation service platform data generation optimization method.

[0014] The present invention also provides a computer-readable storage medium on which executable code is stored. When the executable code is executed by a processor of an electronic device, the processor executes the above-mentioned simulation service platform data generation optimization method.

[0015] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects: The simulation service platform data generation optimization method provided by the present invention can make the calculation result of the compression distance more accurate by optimizing the data generation process, introducing and optimizing the calculation method of the compression distance, making the category discrimination result of the discriminant neural network more accurate, and promoting the generation of higher quality related data by the generation neural network. The method uses two neural network models with complex interactions. Combined with the calculation method of the compression distance, it can simulate complex relationships, effectively solving the problem of the existing data generation method of single model and difficulty in simulating complex relationships.

[0016] The simulation service platform data generation optimization system, equipment and medium provided by the present invention are implemented based on the same technical concept as the above-mentioned scheme and have corresponding technical effects.

[0017] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are only used for the purpose of illustrating specific embodiments and are not to be considered as limiting the present invention. In the entire drawings, the same reference symbols represent the same components; Figure 1 A flowchart of a simulation service platform data generation optimization method provided by an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a simulation service platform data generation optimization system provided by an embodiment of the present invention; Figure 3 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0020] A specific embodiment of the present invention discloses a simulation service platform data generation optimization method, the flow chart is as follows Figure 1 The specific process is described as follows.

[0021] Step S1: The encoder quantizes, encodes and compresses the first input data vector to obtain a first compression distance.

[0022] In this embodiment, the first input data vector may be time series data, sensor data, or other forms of random process samples generated during the operation of the simulation service platform.

[0023] Preferably, before the encoder quantizes the first input data vector, it can also perform: preprocessing operations on the first input data vector. Specifically, the preprocessing operations include but are not limited to normalization, denoising and enhancement. Among them, normalization can be completed by calculating the maximum and minimum values ​​of the input data vector and scaling the input data vector to the interval [0,1] or [-1,1]. Denoising can be performed by methods such as median filtering and Gaussian filtering. Data enhancement can include rotating, scaling, flipping and other operations on the input data vector to increase the diversity of the input data vector. By performing preprocessing operations on the input data vector, the consistency, validity, stability and comparability of the input data vector can be guaranteed. At the same time, by preprocessing the input data vector, this embodiment can make the preprocessed input data vector meet the requirements of subsequent neural network input, and is more suitable for performing subsequent compression distance calculation and category discrimination.

[0024] When the above preprocessing operation is performed before step S1, in step S1, the encoder quantizes, encodes and compresses the preprocessed first input data vector.

[0025] During the quantization process, the encoder can select the quantization factor in the following manner: select a suitable quantization factor based on the distribution characteristics and analysis requirements of the input data vector. The quantization factor can be a fixed value or can be automatically adjusted according to the dynamic range of the input data vector. Quantization and encoding: The encoder multiplies each element in the first input data vector by the quantization factor and rounds it to the nearest integer, and then converts the encoding into a binary string of fixed length. For example, an 8-bit binary number can be used to represent each quantized value, that is, ranging from 0 to 255. In the specific implementation process, a user interface can be provided to the user, allowing the user to adjust the quantization factor according to actual needs to balance data accuracy and compression efficiency. In practical applications, the optimal quantization factor can be determined through experiments so that the quantized data vector can retain sufficient information and effectively reduce the compression length.

[0026] In step S1, after the encoder quantizes and encodes the first input data vector, the quantized encoded first input data vector can be obtained. At this time, the quantized encoded first input data vector is compressed separately, and the size of the compressed file of the quantized encoded first input data vector is recorded to obtain the first compression distance.

[0027] In step S1, the encoder performs quantization, encoding and compression to obtain a compressed distance; the compressed distance, as a low-dimensional representation of high-dimensional data, can be regarded as a feature extraction result of the high-dimensional data. The quantization, encoding and compression process performed by the encoder can realize the conversion of data vectors to distance vectors, thereby extracting the feature representation of the input data.

[0028] Step S2: The discriminant neural network performs category discrimination on the first compression distance to obtain a category discrimination result of the first compression distance.

[0029] The discriminant neural network performs category discrimination on the compression distance. In step S2, the discriminant neural network performs category discrimination on the first compression distance to obtain a category discrimination result of the first compression distance, which is used to indicate the probability that the first compressed data is true data. In one example, when the input data of the discriminant neural network is A, the category discrimination result may be the category of A; when the input data is B, the category discrimination result may be the category of B; when the input data is A and B, the category discrimination result may be whether A and B belong to the same category.

[0030] Combined with the subsequent step S5 for determining the category of the third compressed data, it is possible to determine whether the first compression distance and the third compression distance are of the same category, and obtain a probability value that the third compression distance belongs to the category of the first compression data, thereby reflecting whether the first input data vector and the second input data vector belong to the same category, and whether the second input data generated by the neural network is valid.

[0031] In the discriminant neural network, the relevant category discrimination algorithm can be used to perform category discrimination analysis on the extracted compression distance (as the feature extraction result) to obtain the category discrimination result of the compression distance. In addition, the category discrimination effect can be evaluated by calculating category discrimination indicators (such as silhouette coefficient and Calinski-Harabasz index) to verify the effectiveness of the method of this embodiment in data classification and evaluation.

[0032] Preferably, the discriminant neural network can also be constructed as follows: the discriminant neural network includes an input layer, multiple hidden layers (which can be convolutional layers, pooling layers, etc.) and an output layer. The discriminant neural network can also set an activation layer and use an activation function (such as LeakyReLU or tanh) to increase the nonlinear transformation capability. Finally, the discriminant neural network can also add a fully connected layer and a softmax layer (softmaxLayer or classificationLayer, depending on whether the output probability is required) as the output layer.

[0033] Step S3: Generate a neural network based on the first compression distance and its category discrimination result to generate associated data of the first compression distance; the decoder decompresses and decodes the associated data of the first compression distance to obtain a second input data vector.

[0034] In this embodiment, the neural network can be constructed as follows: an input layer, several fully connected layers (fullyConnectedLayer), each fully connected layer is followed by a ReLU activation layer (reluLayer), and finally the output is reshaped into the desired data size through the reshapeLayer.

[0035] The decoder is connected to the output layer of the generated neural network, and obtains a second input data vector by decompressing and decoding the associated data of the first compression distance.

[0036] Step S4: the encoder quantizes and encodes the second input data vector, and calculates the compression distance between the quantized and encoded first input data vector and the second input data vector to obtain a third compression distance.

[0037] Similar to step S1, before executing step S4, a step of preprocessing the second input data vector may also be added. The specific implementation method is shown in step S1, which will not be described in detail here.

[0038] It is important to ensure that the lengths of the first input data vector and the second input data vector are the same, and that the lengths of the fixed-length binary strings encoded therefrom are the same, so that effective compression distance calculations can be performed later. If the lengths are different, they can be made consistent by interpolation or truncation.

[0039] In step S4, it involves calculating the compression distance between two data vectors, and the specific process is described as follows. The compression distance between the first input data vector and the second input data vector after quantization coding is calculated by executing: Step S41: compress the first input data vector and the second input data vector after quantization encoding separately, and record the size of the compressed files of each separately compressed file.

[0040] Specifically, a data compression algorithm can be used to compress the first input data vector X and the second input data vector Y after quantization encoding into independent data format files, named tempFileX and tempFileY, and the file sizes of the compressed tempFileX and tempFileY (exemplarily, the number of bytes after compression) are recorded respectively, which are expressed as and .

[0041] Common and efficient compression algorithms such as ZIP, LZ77, LZ78, Huffman coding or more advanced algorithms such as Brotli, Zstandard, etc. These algorithms reduce redundant information by identifying repeated patterns and statistical characteristics in the data. You can choose the appropriate compression method by comparing the compression distances under different compression algorithms.

[0042] Step S42: jointly compress the first input data vector and the second input data vector after quantization encoding, and record the size of the jointly compressed compressed file.

[0043] Specifically, merge X and Y into one data set and compress it into tempFileXY using the same compression algorithm. Record the file size of tempFileXY after compression, expressed as .

[0044] Step S43: Calculate the compression distance between the first input data vector and the second input data vector after quantization encoding according to the size of the compressed file of the first input data vector and the second input data vector compressed individually and the size of the compressed file compressed jointly.

[0045] Preferably, in step S42, the first input data vector and the second input data vector after quantization encoding may be concatenated into a long character string; and then, the first input data vector and the second input data vector after quantization encoding may be jointly compressed.

[0046] Preferably, in step S43, the compression distance CD between the first input data vector and the second input data vector after quantization coding is expressed as: , in, , Respectively represent the sizes of compressed files of the first input data vector and the second input data vector after quantization encoding; Indicates the size of the compressed file obtained by jointly compressing the first input data vector and the second input data vector after quantization encoding.

[0047] In this embodiment, the compression distance CD reflects the maximum value of the data format size that is reduced when the two input data vectors are jointly stored compared to separate storage, thereby evaluating the similarity or difference between the input data vectors X and Y. Therefore, the compression distance CD can reflect the information redundancy of X and Y during the compression process, that is, the correlation between them. The smaller the compression distance CD, the higher the information redundancy between X and Y, that is, the stronger the correlation between the two. On the contrary, the larger the compression distance CD, the lower the information redundancy between X and Y, that is, the weaker the correlation between the two.

[0048] Step S5: The discriminant neural network performs category discrimination on the third compression distance to obtain a category discrimination result of the third compression distance; if the category discrimination results of the third compression distance and the first compression distance are the same, the second input data vector is used as the data generation result of the first input data vector; otherwise, jump to the step of generating the associated data of the first compression distance based on the first compression distance and its category discrimination result by the generation neural network, and regenerate the second input data vector.

[0049] In step S5, after the second input data vector is regenerated, the process jumps to step S3 until the category discrimination result of the third compression distance and the first compression distance is one category, and the loop stops.

[0050] It should be noted that in this embodiment, the discriminant neural network and the generative neural network are interrelated and need to be alternately optimized during each iterative training process. In each iterative training process, the parameters of the discriminant neural network are first fixed, and the parameters of the generative neural network are trained to generate more realistic samples; then the parameters of the generative neural network are fixed, and the parameters of the discriminant neural network are trained to more accurately distinguish true and false samples. In the specific implementation process, trainingOptions can be used to set the parameters of the training neural network, such as learning rate, number of iterations, optimization algorithm, etc. Forward propagation and back propagation are performed through forward and backward functions, and network parameters are updated using functions such as adamupdate. The loss function is calculated through forward propagation, and the network parameters are updated using the back propagation algorithm to achieve the goal of the generator generating more realistic samples and the discriminator more accurately distinguishing true and false samples. Through the cyclic iterative process, the parameters of the generative neural network and the discriminant neural network are continuously optimized. In the iterative process, the training of the neural network will make the discrimination distance closer and closer, thereby achieving the goal of the generated data gradually approaching the original data. In the iterative process, by alternately inputting X and Y and other data vectors, it is ensured that the neural network can process data vectors under different data distributions, further improving the diversity and authenticity of the generated data. During the training process, special attention is paid to balancing the training of the generator and the discriminator to avoid problems such as mode collapse and ensure the stability and effectiveness of the training process.

[0051] In summary, the simulation service platform data generation optimization method provided in this embodiment can effectively improve the authenticity and accuracy of the data generated by the generated neural network by optimizing the data generation process. The above method is applicable to both discrete data and continuous data, so it can achieve data generation optimization processing for mixed data. At the same time, by optimizing the quantization factor, compression algorithm and its parameters, the calculation efficiency and accuracy can be effectively improved.

[0052] Embodiment 2 of the present invention discloses a simulation service platform data generation optimization system, the structural diagram of which is shown in FIG. Figure 2 As shown, the system includes an encoder, a discriminant neural network, a generative neural network and a decoder; wherein: The encoder is used to quantize, encode and compress the first input data vector to obtain a first compression distance; and is also used to quantize and encode the second input data vector output by the decoder, and calculate the compression distance between the quantized and encoded first input data vector and the second input data vector to obtain a third compression distance; A discriminant neural network is used to perform category discrimination on the first compression distance to obtain a category discrimination result of the first compression distance; and is also used to perform category discrimination on the third compression distance to obtain a category discrimination result of the third compression distance; if the category discrimination results of the third compression distance and the first compression distance are the same, the second input data vector is used as the data generation result of the first input data vector; otherwise, the generation neural network is instructed to regenerate the associated data of the first compression distance based on the first compression distance and its category discrimination result, until the category discrimination results of the third compression distance and the first compression distance are the same; Generate a neural network for generating associated data of the first compression distance based on the first compression distance and the category discrimination result thereof; The decoder is used to decompress and decode the associated data of the first compression distance to obtain a second input data vector.

[0053] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

[0054] Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 3 , an electronic device includes a memory and a processor.

[0055] 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. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0056] The memory may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage. Among them, ROM may store static data or instructions required by the processor or other modules of the computer. The permanent storage may be a readable and writable storage device. The permanent storage may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at run time. In addition, the memory may include any combination of computer-readable storage media, including various types of semiconductor memory chips (such as DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory may include a readable and / or writable removable storage device, such as a laser disc (CD), a read-only digital versatile disc (such as a DVD-ROM, a double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (such as an SD card, a mini SD card, a Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not include carrier waves and transient electronic signals transmitted wirelessly or wired.

[0057] The memory stores executable codes, and when the executable codes are processed by the processor, the processor can execute part or all of the methods described above.

[0058] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0059] Alternatively, the present application can also be implemented as a computer-readable storage medium (or non-transitory machine-readable storage medium or machine-readable storage medium) on which executable code (or computer program or computer instruction code) is stored. When the executable code (or computer program or computer instruction code) is executed by a processor of an electronic device (or server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0060] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0061] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A simulation service platform data generation optimization method, characterized in that: The method comprises: The encoder quantizes, encodes and compresses the first input data vector to obtain a first compression distance; The discriminant neural network performs category discrimination on the first compression distance to obtain a category discrimination result of the first compression distance; The generating neural network generates associated data of the first compression distance based on the first compression distance and the category discrimination result thereof; the decoder decompresses and decodes the associated data of the first compression distance to obtain a second input data vector; The encoder quantizes and encodes the second input data vector, and calculates a compression distance between the quantized and encoded first input data vector and the second input data vector to obtain a third compression distance; The discriminant neural network performs category discrimination on the third compression distance to obtain a category discrimination result of the third compression distance; if the category discrimination results of the third compression distance and the first compression distance are the same, the second input data vector is used as the data generation result of the first input data vector; otherwise, jump to the step of generating the associated data of the first compression distance based on the first compression distance and its category discrimination result by the generating neural network, and regenerate the second input data vector.

2. The simulation service platform data generation optimization method according to claim 1, characterized in that: The calculation of the compressed distance between the first input data vector and the second input data vector after quantization encoding is performed by: Separately compressing the first input data vector and the second input data vector after quantization encoding, and recording the sizes of the compressed files of the respective compressed files; Combining the first input data vector and the second input data vector after quantization coding, and recording the size of the compressed file of the combined compression; The compression distance between the quantized encoded first input data vector and the second input data vector is calculated according to the sizes of the compressed files of the first input data vector and the second input data vector compressed individually and the size of the compressed file compressed jointly.

3. The simulation service platform data generation optimization method according to claim 2, characterized in that: The compression distance CD between the first input data vector and the second input data vector after quantization coding is expressed as: ， in, , Respectively represent the sizes of compressed files of the first input data vector and the second input data vector after quantization encoding; Indicates the size of the compressed file obtained by jointly compressing the first input data vector and the second input data vector after quantization encoding.

4. The simulation service platform data generation optimization method according to claim 3 is characterized in that: Before jointly compressing the quantized and encoded first input data vector and the second input data vector, the following steps are further performed: Concatenate the quantized and encoded first input data vector and the second input data vector into a long string; Then, the quantized and encoded first input data vector and the second input data vector are jointly compressed.

5. The method for optimizing simulation service platform data generation according to any one of claims 1 to 4, characterized in that: Before the encoder quantizes the first input data vector and the second input data vector, the encoder further performs: Preprocessing operations are performed on the first input data vector and the second input data vector respectively.

6. The simulation service platform data generation optimization method according to claim 5, characterized in that: The preprocessing operations include normalization, denoising and enhancement.

7. The simulation service platform data generation optimization method according to claim 5, characterized in that: The encoder quantizes and encodes the first input data vector and the second input data vector, and performs: The encoder multiplies each element in the first input data vector and the second input data vector by a quantization factor, and rounds to the nearest integer. Then, the encoding is converted into a binary string of fixed length to obtain the first input data vector and the second input data vector after quantization encoding.

8. A simulation service platform data generation optimization system, characterized in that: The system includes an encoder, a discriminative neural network, a generative neural network and a decoder; wherein: The encoder is used to quantize, encode and compress the first input data vector to obtain a first compression distance; and is also used to quantize and encode the second input data vector output by the decoder, and calculate the compression distance between the quantized and encoded first input data vector and the second input data vector to obtain a third compression distance; A discriminant neural network is used to perform category discrimination on the first compression distance to obtain a category discrimination result of the first compression distance; and is also used to perform category discrimination on the third compression distance to obtain a category discrimination result of the third compression distance; if the category discrimination results of the third compression distance and the first compression distance are the same, the second input data vector is used as the data generation result of the first input data vector; otherwise, the generation neural network is instructed to regenerate the associated data of the first compression distance based on the first compression distance and its category discrimination result, until the category discrimination results of the third compression distance and the first compression distance are the same; Generate a neural network for generating associated data of the first compression distance based on the first compression distance and the category discrimination result thereof; The decoder is used to decompress and decode the associated data of the first compression distance to obtain a second input data vector.

9. An electronic device, characterized in that: include: processor; as well as A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the simulation service platform data generation optimization method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: Executable codes are stored thereon, and when the executable codes are executed by a processor of an electronic device, the processor is caused to execute the simulation service platform data generation optimization method as described in any one of claims 1-7.

Images