Process parameter recommendation method and device, chip, medium and program product

Abstract

The embodiments of the present application provide a method and device, chip, medium and program for recommending process parameters under abnormal working conditions. The method includes: generating a first mathematical model based on a first training set; obtaining abnormal process parameters under abnormal working conditions, and performing a recommendation value algorithm using the first mathematical model to obtain training samples of a second training set; converting the production result product indicators of the training samples of the second training set into a multi-objective optimization index; if the multi-objective optimization index is greater than a preset value, optimizing the recommendation value algorithm of the first mathematical model using the second training set; if the multi-objective optimization index is less than a preset value, generating a third training set based on the training samples of the second training set using a production scheduling algorithm; and generating a second mathematical model based on the third training set. The above method can optimize the mathematical model, and when abnormal working conditions occur in the current production, timely adjust the process parameters under the abnormal working conditions, thereby adapting to changes in the abnormal working conditions and meeting the production result indicators of the product.

Description

Technical Field

[0001] This application belongs to the general data processing field of the Internet industry, and in particular relates to a method and device for recommending process parameters for abnormal working conditions, a chip, a medium and a program product. Background Art

[0002] Information technology and production process technologies, represented by artificial intelligence, continue to make significant progress. The use of machine learning algorithms to control and monitor production processes is driving the rapid development of artificial intelligence in the field of production process technology, which is of great significance for controlling production processes and ensuring product quality.

[0003] The production conditions on site are relatively complex, and the usual benchmark operating condition database cannot cover abnormal conditions. For example, due to uncontrollable factors such as equipment load, process parameters such as steam pressure, production gas pressure, and equipment vacuum degree may exceed the pre-set tolerance range of the benchmark operating condition value. If these parameters cannot be restored for a long time, the process parameters will become non-standard operating condition process parameters. At this time, continuing to produce according to the benchmark operating condition process parameters will not achieve the product production result indicators. Summary of the Invention

[0004] The embodiments of the present application provide a method and device, chip, medium and program product for recommending process parameters for abnormal working conditions, which can optimize the mathematical model and, when abnormal working conditions occur in current production, promptly adjust the process parameters under the abnormal working conditions, thereby adapting to changes in the abnormal working conditions and meeting the production result indicators of the product.

[0005] In a first aspect, an embodiment of the present application provides a method for recommending process parameters for abnormal working conditions, comprising:

[0006] generating a first mathematical model based on the first training set;

[0007] Acquire abnormal process parameters under abnormal working conditions, and use the first mathematical model to perform a recommendation value algorithm to obtain training samples of a second training set;

[0008] Converting the production result product indicators of the training samples of the second training set into a multi-objective optimization index;

[0009] If the multi-objective optimization index is greater than a preset value, optimizing the recommendation value algorithm of the first mathematical model using the second training set;

[0010] If the multi-objective optimization index is less than the preset value, generating a third training set based on the training samples of the second training set using a production scheduling algorithm;

[0011] Based on the third training set, a second mathematical model is generated.

[0012] In the above method, the first mathematical model is trained based on the first training set under normal working conditions. When an abnormal working condition occurs, the process parameters under the abnormal working condition are obtained, and the first mathematical model is used to perform a recommendation value algorithm to obtain training samples of the second training set. The resulting product indicators of the training samples of the second training set are then converted into a multi-objective optimization index. If the multi-objective optimization index is greater than a preset value, the second training set is used to optimize the recommendation value algorithm in the first mathematical model. If the multi-objective optimization index cannot reach the preset value, it is necessary to adjust the process parameters of the training samples of the second training set to generate a third training set. Based on the third training set, a second mathematical model is generated and the second mathematical model is used as the final version. This method can optimize the mathematical model and, when abnormal working conditions occur in current production, timely adjust the process parameters under abnormal working conditions to adapt to changes in abnormal working conditions and meet the production result indicators of the product.

[0013] In a possible implementation, generating a third training set by using a production scheduling algorithm based on training samples of the second training set includes:

[0014] Obtaining the number of training samples in the third training set;

[0015] If the number of training samples in the third training set is less than a preset value, the third training set is generated using a production scheduling algorithm.

[0016] In this embodiment of the present application, if the multi-objective optimization index is less than a preset value, the number of training samples in the third training set needs to be determined. If the number is less than the preset value, the number of training samples is insufficient, and the process parameters of the training samples need to be adjusted to generate a third training set until the number of training samples reaches the preset value. This method ensures that the number of training set samples is sufficient, thereby achieving high accuracy in the generated second mathematical model.

[0017] In a possible implementation, if the number of training samples in the third training set is less than a preset value, generating the third training set by using a production scheduling algorithm includes:

[0018] If the number of training samples in the third training set is less than the preset value, adjusting the process parameters of the training samples in the second training set using a production scheduling algorithm to obtain trained samples;

[0019] The third training set is generated according to the trained samples.

[0020] In this embodiment of the present application, if the number of training samples in the third training set is less than a preset value, the production scheduling algorithm is used to adjust the process parameters of the training samples to obtain trained samples, thereby generating a third training set. This method uses the production scheduling algorithm to predict multiple process parameters, ensuring that the number of samples in the third training set meets the preset value standard.

[0021] In a possible implementation, the trained samples include training samples available under abnormal operating conditions and / or training samples unavailable under abnormal operating conditions, and generating the third training set based on the trained samples includes:

[0022] The training samples available under the abnormal working condition are added to the third training set, and the training samples unavailable under the abnormal working condition are deleted from the third training set to obtain the third training set.

[0023] In an embodiment of the present application, trained samples are obtained through production scheduling algorithm training, wherein the trained samples include training samples that are unavailable under abnormal working conditions and / or training samples that are available under abnormal working conditions. The unavailable training samples under abnormal working conditions are eliminated, and the available training samples under abnormal working conditions are added to the third training set to obtain the third training set, thereby ensuring the accuracy and effectiveness of the third training set.

[0024] In a possible implementation, generating the first mathematical model based on the first training set includes:

[0025] Under normal operating conditions, the first training set is generated using product design parameters based on the design scorecard in the R&D stage;

[0026] The first mathematical model is generated based on the first training set. In an embodiment of the present application, the first mathematical model is obtained by training the first training set under normal operating conditions, wherein the first training set is generated under normal operating conditions based on a design scorecard in the research and development phase and using product design parameters. However, when abnormal operating conditions occur, the first mathematical model will not meet the requirements of actual production, and the first mathematical model needs to be optimized to adapt to changes in the abnormal operating conditions and meet the production result indicators of the product.

[0027] In a second aspect, an embodiment of the present application provides a process parameter recommendation device for abnormal working conditions, comprising:

[0028] A processing unit, configured to generate a first mathematical model based on a first training set;

[0029] an acquisition unit, configured to acquire abnormal process parameters under abnormal working conditions, and perform a recommendation value algorithm using the first mathematical model to obtain training samples of a second training set;

[0030] The processing unit is used to convert the production result product indicators of the training samples of the second training set into a multi-objective optimization index; if the multi-objective optimization index is greater than a preset value, the second training set is used to optimize the recommendation value algorithm of the first mathematical model; if the multi-objective optimization index is less than the preset value, a third training set is generated based on the training samples of the second training set using a production scheduling algorithm; and a second mathematical model is generated based on the third training set.

[0031] The operations and beneficial effects performed by the process parameter recommendation device for abnormal working conditions can refer to the methods and beneficial effects described in any one of the first aspects above, and the repeated parts will not be repeated.

[0032] In the third aspect, the present application provides a device for recommending process parameters for abnormal working conditions, which may be a server, a device in a server, or a device that can be used in conjunction with a server. The device for recommending process parameters for abnormal working conditions may also be a chip system. The device for recommending process parameters for abnormal working conditions may execute any of the methods described in the first aspect. The functions of the device for recommending process parameters for abnormal working conditions may be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The module may be software and / or hardware. The operations and beneficial effects performed by the device for recommending process parameters for abnormal working conditions may refer to the methods and beneficial effects described in any of the first aspects above, and the repeated parts will not be repeated.

[0033] In a fourth aspect, the present application provides a device for recommending process parameters for abnormal working conditions, wherein the device for recommending process parameters for abnormal working conditions comprises a processor. When the processor calls a computer program in a memory, the method described in any one of the first aspects is executed.

[0034] In a fifth aspect, the present application provides a device for recommending process parameters for abnormal working conditions, the device comprising a processor and a memory, the memory being used to store a computer program; the processor being used to execute the computer program stored in the memory, so that the device for recommending process parameters for abnormal working conditions executes the method described in any one of the first aspects.

[0035] In the sixth aspect, the present application provides a device for recommending process parameters for abnormal working conditions, which includes a processor, a memory and a transceiver, wherein the transceiver is used to receive channels or signals, or send channels or signals; the memory is used to store computer programs; and the processor is used to call the computer program from the memory to execute the method described in any one of the first aspects.

[0036] In the seventh aspect, the present application provides a process parameter recommendation device for abnormal working conditions, which includes a processor and an interface circuit, wherein the interface circuit is used to receive a computer program and transmit it to the processor; the processor runs the computer program to execute the method as described in any one of the first aspects.

[0037] In an eighth aspect, the present application provides a computer-readable storage medium for storing a computer program, which, when executed, enables the method described in any one of the first aspects to be implemented.

[0038] In a ninth aspect, the present application provides a computer program product comprising a computer program, which, when executed, enables the method described in any one of the first aspects to be implemented.

[0039] In the tenth aspect, an embodiment of the present application provides a process parameter recommendation system for abnormal working conditions, which includes at least one server and at least one terminal device, and the server is used to execute any one of the steps in the above-mentioned first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The following is an introduction to the drawings used in the embodiments of this application.

[0041] Figure 1 This is a schematic diagram of the architecture of a process parameter recommendation system for abnormal working conditions provided by an embodiment of the present application;

[0042] Figure 2 is a schematic diagram of a recurrent neural network 200 provided in an embodiment of the present application;

[0043] Figure 3 This is a flow chart of a method for recommending process parameters for abnormal working conditions provided in an embodiment of the present application;

[0044] Figure 4 4 is a schematic structural diagram of a process parameter recommendation device 400 for abnormal working conditions provided in an embodiment of the present application;

[0045] Figure 5 2 is a schematic structural diagram of a process parameter recommendation device 500 for abnormal working conditions provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The embodiments of the present application are described below in conjunction with the drawings in the embodiments of the present application.

[0047] Since the embodiments of the present application involve the application of a large number of neural networks, in order to facilitate understanding, the relevant terms and related concepts such as neural networks involved in the embodiments of the present application are first introduced below.

[0048] (1) Neural Network

[0049] A neural network can be composed of neural units, which can be represented by x s The output of the operation unit with the intercept 1 as input can be:

[0050]

[0051] Where, s = 1, 2, ... n, n is a natural number greater than 1, W s is x s The weight of the neural unit, b is the bias of the neural unit. f is the activation function of the neural unit, which is used to introduce nonlinear characteristics into the neural network to convert the input signal in the neural unit into the output signal. The output signal of the activation function can be used as the input of the next convolutional layer. The activation function can be a sigmoid function. A neural network is a network formed by connecting many of the above-mentioned single neural units together, that is, the output of one neural unit can be the input of another neural unit. The input of each neural unit can be connected to the local receptive field of the previous layer to extract the features of the local receptive field. The local receptive field can be an area composed of several neural units.

[0052] (2) Deep Neural Networks

[0053] A deep neural network (DNN), also known as a multi-layer neural network, can be understood as a neural network with many hidden layers. The term "many" here doesn't have a specific metric. Based on the location of different layers within a DNN, the neural network can be divided into three categories: input layer, hidden layer, and output layer. Generally speaking, the first layer is the input layer, the last layer is the output layer, and all layers in between are hidden layers. Layers are fully connected, meaning that any neuron in layer i is connected to any neuron in layer i+1. While DNNs appear complex, the operation of each layer is actually quite simple. Simply put, it's the following linear relationship: y = α(Wx + b), where x is the input vector, y is the output vector, b is the bias vector, W is the weight matrix (also known as coefficient), and α() is the activation function. Each layer simply performs this simple operation on the input vector x to produce the output vector y. Due to the large number of DNN layers, the number of coefficients W and bias vectors b is also large. The definition of these parameters in DNN is as follows: Take the coefficient W as an example: Assume that in a three-layer DNN, the linear coefficient from the 4th neuron in the second layer to the 2nd neuron in the third layer is defined as The superscript 3 represents the layer number of the coefficient W, while the subscript corresponds to the output of the third layer index 2 and the input of the second layer index 4. In summary, the coefficient from the kth neuron in the L-1th layer to the jth neuron in the Lth layer is defined as It's important to note that the input layer has no W parameter. In deep neural networks, more hidden layers allow the network to better capture complex real-world situations. Theoretically, a model with more parameters has higher complexity and greater "capacity," meaning it can handle more complex learning tasks. Training a deep neural network is essentially the process of learning the weight matrix, with the ultimate goal of obtaining the weight matrices for all layers of a trained deep neural network (a weight matrix formed by the vectors W across many layers).

[0054] (3) Convolutional Neural Network

[0055] A convolutional neural network (CNN) is a deep neural network with a convolutional architecture. A CNN consists of a feature extractor consisting of convolutional layers and subsampling layers. This feature extractor can be viewed as a filter, and the convolution process can be thought of as convolving an input image or feature map with a trainable filter. A convolutional layer is the layer of neurons in a CNN that performs convolution on the input signal. Within a convolutional layer, a neuron can only connect to a subset of neurons in adjacent layers. A convolutional layer typically contains several feature planes, each composed of a rectangular arrangement of neurons. Neurons within the same feature plane share weights, referred to as the convolution kernel. Shared weights can be understood as ensuring that the method for extracting image information is independent of position. The underlying principle is that the statistical information of one part of an image is the same as that of another. This means that image information learned in one part can also be applied to other parts. Therefore, the same learned image information is available at all positions in the image. In the same convolutional layer, multiple convolution kernels can be used to extract different image information. Generally speaking, the more convolution kernels there are, the richer the image information reflected by the convolution operation.

[0056] Convolution kernels can be initialized as matrices of random size, and during the training process of the convolutional neural network, the convolution kernels can be learned to obtain reasonable weights. In addition, the direct benefit of shared weights is that they reduce the number of connections between the layers of the convolutional neural network, while also reducing the risk of overfitting.

[0057] (4) Recurrent Neural Networks (RNN) are used to process sequence data. In traditional neural network models, the layers are fully connected from the input layer to the hidden layer and then to the output layer, while the nodes within each layer are disconnected. Although this ordinary neural network solves many difficult problems, it is still powerless to solve many problems. For example, if you want to predict the next word in a sentence, you generally need to use the previous word because the previous and next words in a sentence are not independent. The reason why RNN is called a recurrent neural network is that the current output of a sequence is also related to the previous output. The specific manifestation is that the network will remember the previous information and apply it to the calculation of the current output, that is, the nodes between the hidden layers are no longer disconnected but connected, and the input of the hidden layer includes not only the output of the input layer but also the output of the hidden layer at the previous moment. In theory, RNN can process sequence data of any length. The training of RNN is the same as the training of traditional CNN or DNN. This approach also uses the backpropagation algorithm, but with one key difference: if the RNN is expanded, its parameters, such as W, are shared; this is not the case with traditional neural networks, as in the example above. Furthermore, when using gradient descent, the output of each step depends not only on the state of the network at the current step but also on the state of the network at several previous steps. This learning algorithm is called backpropagation through time (BPTT).

[0058] Given the existence of convolutional neural networks, the reason for recurrent neural networks is simple. Convolutional neural networks assume that elements are independent of each other, and that inputs and outputs are also independent, such as cats and dogs. However, in the real world, many elements are interconnected, such as the changes in stock prices over time. Or, for example, someone says, "I love traveling, and my favorite place is Yunnan. I must visit it someday." Humans should know to fill in the blank with "Yunnan." Because humans infer context, RNNs were created to enable machines to do this. RNNs aim to give machines human-like memory. Therefore, the output of an RNN depends on both current input and historical memory.

[0059] See Figure 1 , Figure 1This is a schematic diagram of the architecture of a process parameter recommendation system for abnormal working conditions provided in an embodiment of the present application. As shown in the figure, the data acquisition device 106 is used to obtain abnormal process parameters under abnormal working conditions. In the embodiment of the present application, the data includes: product design parameters and abnormal process parameters under normal working conditions; and the abnormal process parameters are stored in the database 108. The training device 109 obtains the target model / rule 101 based on the abnormal process parameters in the database 108, wherein the target model / rule 101 can be a mathematical model. The following will describe in more detail how the training device 109 obtains the target model / rule 101 based on the training data set. The target model / rule 101 can be used to implement the process parameter recommendation method for abnormal working conditions provided in an embodiment of the present application. The target model / rule 101 in the embodiment of the present application can specifically be a recurrent neural network. In the embodiment provided in the present application, the recurrent neural network is obtained by training the model to be trained. It should be noted that in actual applications, the data in the database 108 does not necessarily come from the collection of the data acquisition device 106, but may also be received from other devices. It should also be noted that the training device 109 does not necessarily train the target model / rule 101 entirely based on the training data set of the database 108. It is also possible to obtain the training data set from the cloud or other places for model training. The above description should not be used as a limitation on the embodiments of the present application.

[0060] The target model / rule 101 obtained by training the training device 109 can be applied to different systems or devices, such as Figure 1 The execution device 110 shown in the figure can be a terminal, such as a mobile phone terminal, a tablet computer, a laptop computer, an augmented reality / virtual reality (AR / VR), a vehicle terminal, etc. It can also be a server or a cloud. Figure 1 In the embodiment, the execution device 110 is configured with an I / O interface 102 for data interaction with an external device. A user can input data into the I / O interface 102 through a client device 104 .

[0061] The preprocessing module 103 is used to preprocess the abnormal process parameters under abnormal working conditions received by the I / O interface 102. In an embodiment of the present application, the preprocessing module 103 can be used to obtain training samples of the second training set based on the abnormal process parameters under abnormal working conditions using a recommendation value algorithm, thereby constraining the target model / rule 101.

[0062] When the execution device 110 preprocesses the training samples under abnormal working conditions, or when the calculation module 107 of the execution device 110 performs calculations and other related processing, the execution device 110 can call the data, code, etc. in the data storage system 105 for corresponding processing, and can also store the data, instructions, etc. obtained from the corresponding processing in the data storage system 105.

[0063] Finally, the I / O interface 102 returns the processed result to the client device 104 so as to provide it to the user.

[0064] It is worth noting that the training device 109 can generate corresponding target models / rules 101 based on different training data for different goals or different tasks. The corresponding target models / rules 101 can be used to achieve the above goals or complete the above tasks, thereby providing users with the desired results.

[0065] exist Figure 1 In the illustrated scenario, the user can manually input data, which can be performed through the interface provided by I / O interface 102. In another scenario, client device 104 can automatically send input data to I / O interface 102. If user authorization is required for client device 104 to automatically send input data, the user can set the corresponding permissions in client device 104. The user can view the results output by execution device 110 on client device 104, which can be presented in the form of a display, sound, action, or other specific methods. Client device 104 can also serve as a data acquisition terminal, collecting input data input into I / O interface 102 and output results output from I / O interface 102 as new sample data and storing them in database 108. Of course, collection can also be performed without client device 104, with I / O interface 102 directly storing the input data input into I / O interface 102 and output results output from I / O interface 102 as new sample data in database 108.

[0066] It is worth noting that Figure 1 This is only a schematic diagram of a system architecture provided by an embodiment of the present invention. The positional relationship between the devices, components, modules, etc. shown in the figure does not constitute any limitation. For example, Figure 1 In the embodiment, the data storage system 105 is an external memory relative to the execution device 110. In other cases, the data storage system 105 can also be placed in the execution device 110.

[0067] like Figure 1 As shown, the target model / rule 101 is obtained through training by the training device 109. The target model / rule 101 can be a recurrent neural network in the embodiment of the present application.

[0068] See Figure 2 , Figure 2 2 is a schematic diagram of a recurrent neural network 200 provided in an embodiment of the present application, which may include an input layer 210, a convolutional layer / pooling layer 220, and a neural network layer 230. The convolutional layer / pooling layer 220 may include layers 221-226. For example, in one implementation, layer 221 is a convolutional layer, layer 222 is a pooling layer, layer 223 is a convolutional layer, layer 224 is a pooling layer, layer 225 is a convolutional layer, and layer 226 is a pooling layer. In another implementation, layers 221 and 222 are convolutional layers, layer 223 is a pooling layer, layer 224 and layer 225 are convolutional layers, and layer 226 is a pooling layer. That is, the output of a convolutional layer can be used as the input of a subsequent pooling layer, or as the input of another convolutional layer to continue the convolution operation.

[0069] The following will take convolutional layer 221 as an example to introduce the internal working principle of a convolutional layer.

[0070] The convolution layer 221 can include multiple convolution operators, also known as kernels. During model training, a convolution operator acts as a filter that extracts specific information from the input matrix. Essentially, a convolution operator is a weight matrix, which is typically predefined. During the convolution operation on the input matrix, the weight matrix typically processes the input matrix vector horizontally, one eigenvalue after another or two eigenvalues ​​after two eigenvalues, to extract specific features from the matrix. The size of the weight matrix should be related to the size of the matrix. It is important to note that the depth dimension of the weight matrix is ​​the same as the depth dimension of the input matrix. During the convolution operation, the weight matrix extends to the entire depth of the input matrix. Therefore, convolution with a single weight matrix produces a convolution output with a single depth dimension. However, in most cases, a single weight matrix is ​​not used. Instead, multiple weight matrices of the same size (row × column) are applied, i.e., multiple homogeneous matrices. The outputs of each weight matrix are stacked to form the depth dimension of the convolved image, where the dimension can be understood as being determined by the "multiple" mentioned above. The multiple weight matrices have the same size (rows × columns), and the feature maps extracted by the multiple weight matrices of the same size are also of the same size. The multiple features of the same size extracted are then merged to form the output of the convolution operation.

[0071] The weight values ​​in these weight matrices need to be obtained through a lot of training in practical applications. The weight matrices formed by the weight values ​​obtained through training can be used to extract information from the input matrix, so that the recurrent neural network 200 can make correct predictions.

[0072] When the recurrent neural network 200 has multiple convolutional layers, the initial convolutional layer (for example, 221) often extracts more general features, which can also be called low-level features. As the depth of the recurrent neural network 200 increases, the features extracted by the subsequent convolutional layers (for example, 226) become more and more complex, such as high-level semantic features. Features with higher semantics are more suitable for the problem to be solved.

[0073] Since it is often necessary to reduce the number of training parameters, it is often necessary to periodically introduce a pooling layer after the convolution layer, such as Figure 2 The layers 221-226 illustrated in the convolutional layer / pooling layer 220 may be a convolutional layer followed by a pooling layer, or multiple convolutional layers may be followed by one or more pooling layers.

[0074] After being processed by the convolution layer / pooling layer 220, the recurrent neural network 200 is not sufficient to output the required output information. As mentioned above, the convolution layer / pooling layer 220 only extracts features. However, in order to generate the final output information (the required class information or other related information), the recurrent neural network 200 needs to use the neural network layer 230 to generate one or a group of outputs of the required number of classes. Therefore, the neural network layer 230 may include multiple hidden layers (such as Figure 2 231, 232 to 23n) and the output layer 240 shown, the parameters contained in the multi-layer hidden layer can be pre-trained according to relevant training data of a specific task type.

[0075] After the multiple hidden layers in the neural network layer 230, that is, the last layer of the entire recurrent neural network 200 is the output layer 240, which has a loss function similar to the classification cross entropy, specifically used to calculate the prediction error. Once the forward propagation of the entire recurrent neural network 200 (such as Figure 2 The propagation from 210 to 240 is forward propagation) and the reverse propagation (such as Figure 2 The propagation from 240 to 210 is called back propagation) and the weight values ​​and biases of the aforementioned layers will begin to be updated to reduce the loss of the recurrent neural network 200 and the error between the result output by the recurrent neural network 200 through the output layer and the ideal result.

[0076] It should be noted that Figure 2 The shown recurrent neural network 200 is only an example of a recurrent neural network. In specific applications, the recurrent neural network may also exist in the form of other network models.

[0077] See Figure 3 , Figure 3This is a flow chart of a method for recommending process parameters for abnormal working conditions provided by an embodiment of the present application, which includes but is not limited to the following steps:

[0078] S301: Generate a first mathematical model based on a first training set.

[0079] In some embodiments, under normal operating conditions, a first training set is generated using product design parameters based on a design scorecard in the R&D phase; and a first mathematical model is generated based on the first training set.

[0080] Specifically, normal operating conditions, namely standard operating conditions, refer to the working conditions of production equipment under standard conditions, wherein the operating conditions are determined based on multiple process parameters. For example, for the process parameter of production temperature, the temperature range within interval A is one operating condition, and the temperature range within interval B (B is different from A) is another operating condition. Each process parameter corresponds to a benchmark operating condition range (the range of process parameters under standard operating conditions). If the process parameter exceeds the tolerance range preset in the benchmark operating condition range and cannot be restored for a long time, the process parameter will become a non-standard operating condition parameter, that is, an abnormal operating condition parameter. The above-mentioned product design parameters may include controllable parameter indicators of different production equipment, wherein the controllable parameter indicators include at least one of temperature, humidity, steam pressure, production gas pressure, and equipment vacuum. In this embodiment, the first mathematical model is obtained by training a first training set under normal operating conditions, wherein the first training set is generated under normal operating conditions using product design parameters based on a design scorecard in the research and development stage. However, when abnormal operating conditions occur, the first mathematical model will not be able to meet the actual production requirements, and the first mathematical model needs to be optimized to adapt to changes in abnormal operating conditions and meet the production result indicators of the product.

[0081] S302: Acquire abnormal process parameters under abnormal working conditions, perform a recommendation value algorithm using the first mathematical model, and obtain training samples of a second training set.

[0082] Specifically, abnormal process parameters under abnormal working conditions are obtained, and a recommended value algorithm is performed on the abnormal process parameters using the first mathematical model to obtain training samples for the second training set. When abnormal production working conditions occur during the production process, the process parameters under the abnormal working conditions are monitored, and information such as the non-standard working condition code, the current production batch number, and the time the benchmark working condition has been responded to are recorded. At the same time, the process parameters under the abnormal working conditions and the benchmark working condition range of the process parameters are obtained.

[0083] For example, there are P1 to Pn process flows, each flow has m process parameters X, which can be expressed as Pn: X1 to Xm, and the operating parameter fault codes are: N, M: A1, B1, C1, D1, E1, ···Ai, Bi, Ci, Di, Ei (N is the number of abnormal processes, M is the number of abnormal parameters X, A is the process number, B is the process parameter number, C is the process parameter fault code, D is the fault code available flag, and E is the abnormal parameter database address number).

[0084] S303: Convert the production result product indicators of the training samples of the second training set into multi-objective optimization indices.

[0085] Specifically, the first mathematical model is called, and other recommended process parameters corresponding to the abnormal process parameters are calculated using the tuning algorithm. Actual production is carried out according to the recommended process parameters, thereby obtaining a multi-objective optimization index.

[0086] The above-mentioned multi-objective optimization index refers to the evaluation score of the target product, which can be represented by the letter P, where P ranges from 0 to 100, with a full score of 100. The larger P is, the more accurate the mathematical model is.

[0087] After obtaining the process parameters under abnormal working conditions, the process parameters X=(X1, X2, ..Xm) are input into the first mathematical model, and the actual value of the product parameter Y=(Y1, Y2, ..Yn) is obtained by setting it as the equipment process parameter through production scheduling and production verification. Then, based on the actual value of the product parameter Y, a multi-objective optimization index P=F(Y target value, Y actual value) is generated, where the Y target value is the product parameter target value set by the multi-objective machine learning algorithm. The target value is comprehensively set in combination with the customer's minimum required quality of the product parameter, the specification limit, and the fluctuation of the product parameter in mass production. To ensure a 100% product yield, for example, a customer may require a primary amine content of 99%. Therefore, a product with a primary amine content below 99% is considered defective. Because the product parameters of these products follow a normal distribution, if production is conducted according to the 99% standard, there is a 50% chance that the product will eventually fall below 99%, resulting in a defective product. Typically, the target value is set to the lower specification limit plus greater than four times the standard deviation to ensure a zero probability of the product parameter falling below 99%. For example, if the standard deviation of the primary amine content is 0.1%, the target value Y is set to be greater than or equal to 99.4%. The actual value Y is the actual value of the product parameter under abnormal operating conditions.

[0088] S304: If the multi-objective optimization index is greater than a preset value, the recommendation value algorithm of the first mathematical model is optimized using the second training set.

[0089] Specifically, if the multi-objective optimization index is greater than a preset value, the second training set obtained by performing the recommendation value algorithm using the first mathematical model is used to optimize the recommendation value algorithm of the first mathematical model.

[0090] The above-mentioned second training set is established based on the non-standard working condition coding, wherein the format of the training sample database in the training set is: number, non-standard working condition coding, production batch, benchmark working condition values ​​of the process parameters of the entire process P1~Pn, recommended parameter values, and corresponding product quality characteristic values.

[0091] The production data corresponding to the production batch can be saved via blockchain based on the production batch. A first blockchain address can be generated for the first blockchain of the production data corresponding to the current production batch, and then a second blockchain address can be generated for the second blockchain storing the data in the second training set. The production data corresponding to the current production batch can then be verified to determine whether to add it to the second training set. If it is added to the second training set, the production data is obtained based on the first blockchain address, and the corresponding content in the second blockchain is added to the first blockchain based on the second blockchain address. If it is deleted, the data of the current batch included in the second blockchain is deleted based on the second blockchain address.

[0092] S305: If the multi-objective optimization index is less than a preset value, a third training set is generated using a production scheduling algorithm based on the training samples of the second training set.

[0093] In some embodiments, if the multi-objective optimization index is less than the preset value, the number of training samples of the third training set is obtained; if the number of training samples of the third training set is less than the preset value, the third training set is generated using a production scheduling algorithm.

[0094] Furthermore, if the number of training samples in the third training set is less than a preset value, the process parameters of the training samples in the second training set are adjusted through the production scheduling algorithm to obtain trained samples, wherein the trained samples include training samples available under abnormal working conditions and / or training samples unavailable under abnormal working conditions, the training samples available under abnormal working conditions are added to the third training set, and the training samples unavailable under abnormal working conditions are deleted from the third training set to obtain the third training set.

[0095] Specifically, obtain the number of training samples in the third training set. When the number of training samples in the third training set is less than the preset value, at this time, the training samples are insufficient, and it is necessary to continue learning through the training scheduling recommendation method to obtain the trained samples. Then, encode the working conditions of all abnormal process parameters produced in the abnormal batches as the process parameters of this available batch and add them to the third training set. At the same time, remove the process parameters that do not meet the overall requirements in the training set (the mean and number of changes of the process parameters of the training samples exceed the preset range, the data quality of the training samples is abnormal, etc.), that is, the process parameters that are unavailable under abnormal working conditions are removed from the third training set, and finally, the third training set is obtained. Among them, the specific implementation method of the training scheduling recommendation method is as follows:

[0096] Compare the multi-objective optimization index P values of the two most recent abnormal trainings. When P(K + i)>P(K), where K is the number of training rounds, K >= 1 and K < N (N is a positive integer), use the nearby interval corresponding to P(K + i) of the process parameter Xn for new training scheduling recommendations.

[0097] For example, the initial recommended value obtained by the algorithm for the process parameter Xn of the training samples in the second training set obtained by the abnormal working condition recommendation algorithm in this round is 1100, and P(K)=75.

[0098] (1) Generate an interval of 1100 + / - 50 near the process parameter Xn, and randomly select the parameter for the next round as 1075 from several values (1050, 1075, 1125, 1150);

[0099] (2) If the recommended value for the next round K is in the training stage, then first use 1075 for the process parameter Xn, P(K + 1)=74, and P(K + 1)<P(K)=75 in this round, so do not continue to recommend using the Xn value in this round;

[0100] (3) The recommended value for the K + 2 round (randomly recommend 1150), Xn = 1150, P(K + 2)=80, and P(K + 2)>P(K) in this round. Generate an interval of 1150 + / - 50 near Xn = 1150, and randomly select data from the interval values (1100, 1125, 1150, 1175, 1200) for the recommendation algorithm. For example, select the recommended value as Xn = 1175;

[0101] (4) The recommended value for the K + 3 round is Xn = 1175, P(K + 3)=82, and P(K + 3)>P(K + 2) in this round. Generate an interval value of 1175 + / - 50 near Xn = 1175, randomly select data for the recommendation algorithm, and the subsequent recommendations are carried out in the same way.

[0102] Add the process parameters generated in each round to the third training set until the number of training samples in the third training set is greater than the preset value.

[0103] It should be noted that the training flag is set to ensure that the process parameter Xn of the training sample of the third training set recommended each time fluctuates randomly around the recommended value of the previous round. If the multi-objective optimization index P value of the recommendation result of this round is good, the recommended parameters will continue to be generated near the Xn parameter value of this round. Through this training algorithm, the improvement of the multi-objective optimization index P value of each recommendation algorithm can be intuitively compared, while also meeting the needs of training samples.

[0104] In this embodiment, if the multi-objective optimization index is less than the preset value, it means that the above-mentioned first mathematical model fails and cannot meet the actual production under abnormal working conditions. At this time, it is necessary to train the production scheduling algorithm, adjust the process parameters of the training samples of the second training set, and generate a third training set to prepare for the subsequent generation of the second mathematical model.

[0105] S306: Generate a second mathematical model based on the third training set.

[0106] Specifically, the training samples of the third training set include process parameters obtained through the production scheduling algorithm and a sufficient number of training samples obtained through production verification. The third training set is used to generate a second mathematical model through machine learning methods, and the trained second mathematical model is used as the final version.

[0107] Furthermore, the above-mentioned second mathematical model is used to perform a recommendation value algorithm to obtain training samples of the fourth training set, and then the production result product indicators of the training samples of the fourth training set are converted into a second multi-objective optimization index (the specific implementation method can refer to the specific implementation method of steps S302-S303, which will not be described in detail here), until the second multi-objective optimization index is greater than the preset value, and the recommended value of the process parameter Xn of the training sample of the fourth training set that finally meets the standard is set as the final recommended value of the abnormal working condition, and used under the abnormal working condition.

[0108] The above describes in detail the method of the embodiment of the present application, and the following provides an apparatus of the embodiment of the present application.

[0109] See Figure 4 , Figure 4 4 is a structural diagram of a process parameter recommendation device 400 for abnormal working conditions provided in an embodiment of the present application. The device includes a processing unit 401 and an acquisition unit 402, wherein each unit is described in detail as follows.

[0110] The processing unit 401 is configured to generate a first mathematical model based on a first training set;

[0111] An acquisition unit 402 is configured to acquire abnormal process parameters under abnormal working conditions, and perform a recommendation value algorithm using the first mathematical model to obtain training samples of a second training set;

[0112] The processing unit 401 is used to convert the production result product indicators of the training samples of the second training set into a multi-objective optimization index; if the multi-objective optimization index is greater than a preset value, the second training set is used to optimize the recommendation value algorithm of the first mathematical model; if the multi-objective optimization index is less than the preset value, a third training set is generated based on the training samples of the second training set using a production scheduling algorithm; and a second mathematical model is generated based on the third training set.

[0113] In a possible implementation, the acquiring unit 402 is further configured to:

[0114] Obtaining the number of training samples in the third training set;

[0115] The processing unit 401 is specifically configured to:

[0116] If the number of training samples in the third training set is less than a preset value, the third training set is generated using a production scheduling algorithm.

[0117] In a possible implementation, the processing unit 401 is specifically configured to:

[0118] If the number of training samples in the third training set is less than the preset value, adjusting the process parameters of the training samples in the second training set using a production scheduling algorithm to obtain trained samples;

[0119] The third training set is generated according to the trained samples.

[0120] In a possible implementation, the trained samples include training samples available under abnormal operating conditions and / or training samples unavailable under abnormal operating conditions, and the processing unit 401 is specifically configured to:

[0121] The training samples available under the abnormal working condition are added to the third training set, and the training samples unavailable under the abnormal working condition are deleted from the third training set to obtain the third training set.

[0122] In a possible implementation, the processing unit 401 is further configured to:

[0123] Under normal operating conditions, the first training set is generated using product design parameters based on the design scorecard in the R&D stage;

[0124] Based on the first training set, the first mathematical model is generated. It should be noted that the implementation and beneficial effects of each unit can also refer to Figure 3 The corresponding description of the method embodiment shown.

[0125] See Figure 5 , Figure 5 This is a structural diagram of a process parameter recommendation device 500 for abnormal working conditions provided in an embodiment of the present application. The device 500 includes a processor 501 and a transceiver 503, and optionally, also includes a memory 502. The processor 501, memory 502 and transceiver 503 are interconnected via a bus 504.

[0126] The memory 502 includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or compact disc read-only memory (CD-ROM), and is used for related instructions and data. The transceiver 503 is used to receive and send data.

[0127] The processor 501 may be one or more central processing units (CPUs). In the case where the processor 501 is a CPU, the CPU may be a single-core CPU or a multi-core CPU.

[0128] The processor 501 in the apparatus 500 reads the program code stored in the memory 502 to perform the following operations:

[0129] generating a first mathematical model based on the first training set;

[0130] Acquire abnormal process parameters under abnormal working conditions, and use the first mathematical model to perform a recommendation value algorithm to obtain training samples of a second training set;

[0131] Converting the production result product indicators of the training samples of the second training set into a multi-objective optimization index;

[0132] If the multi-objective optimization index is greater than a preset value, optimizing the recommendation value algorithm of the first mathematical model using the second training set;

[0133] If the multi-objective optimization index is less than the preset value, generating a third training set based on the training samples of the second training set using a production scheduling algorithm;

[0134] Based on the third training set, a second mathematical model is generated.

[0135] Optionally, the number of training samples in the third training set is obtained through the transceiver 503, and the processor 501 is further configured to generate the third training set using a scheduling algorithm if the number of training samples in the third training set is less than a preset value.

[0136] Optionally, the processor 501 is also used to adjust the process parameters of the training samples of the second training set using a production scheduling algorithm to obtain trained samples if the number of training samples in the third training set is less than the preset value; and generate the third training set based on the trained samples.

[0137] Optionally, the trained samples include training samples available under abnormal working conditions and / or training samples unavailable under abnormal working conditions. The processor 501 is further used to add the training samples available under abnormal working conditions to the third training set, and delete the training samples unavailable under abnormal working conditions from the third training set to obtain the third training set.

[0138] Optionally, the processor 501 is further configured to generate, under normal operating conditions, the first training set using product design parameters based on a design scorecard in the R&D stage; and generate the first mathematical model based on the first training set.

[0139] It should be noted that the implementation and beneficial effects of each operation can also refer to Figure 3 The corresponding description of the method embodiment shown.

[0140] The present application also provides a chip system, which includes a processor for supporting a server to implement the functions involved in any of the above embodiments. In one possible design, the chip system may also include a memory for locating computer programs and data necessary for the server. The chip system can be composed of a chip or can include a chip and other discrete devices. The input and output of the chip system correspond to the receiving and sending operations of the server in the method embodiment, respectively.

[0141] The present application also provides a device for recommending process parameters for abnormal working conditions, including a processor and an interface. The processor can be used to execute the method in the above method embodiment.

[0142] It should be understood that the process parameter recommendation device for abnormal operating conditions may be a chip. For example, the process parameter recommendation device for abnormal operating conditions may be a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on chip (SoC), a central processor unit (CPU), a network processor (NP), a digital signal processor (DSP), a microcontroller unit (MCU), a programmable logic device (PLD), or other integrated chips.

[0143] During implementation, each step of the above method can be completed by an integrated logic circuit of the hardware in the processor or by instructions in the form of software. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly embodied as being executed by a hardware processor, or can be executed by a combination of hardware and software modules in the processor. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in conjunction with its hardware. To avoid repetition, it will not be described in detail here.

[0144] It should be noted that the processor in the embodiments of the present application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiment can be completed by an integrated logic circuit of the hardware in the processor or by instructions in the form of software. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiments of the present application can be directly embodied as being executed by a hardware decoding processor, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0145] According to the method provided in the embodiment of the present application, the present application also provides a computer program product, which includes: a computer program, when the computer program is run on a computer, causes the computer to execute Figure 3 A method according to any one of the embodiments shown.

[0146] According to the method provided in the embodiment of the present application, the present application also provides a computer readable medium, which stores a computer program, which, when executed on a computer, causes the computer to execute Figure 3 A method according to any one of the embodiments shown.

[0147] According to the method provided in the embodiment of the present application, the present application also provides a process parameter recommendation system for abnormal working conditions, which includes the aforementioned one or more servers and one or more terminal devices.

[0148] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated therein. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a high-density digital video disc (DVD)), or a semiconductor medium (eg, a solid state disc (SSD)).

[0149] Those skilled in the art will appreciate that the various illustrative logical blocks and steps described in conjunction with the embodiments disclosed herein can be implemented using electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0150] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0151] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a positioning server, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0152] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for recommending process parameters for abnormal working conditions, characterized in that: include: generating a first mathematical model based on the first training set; The first training set is generated under normal operating conditions using product design parameters based on a design scorecard from the R&D phase; Acquire abnormal process parameters under abnormal working conditions, and use the first mathematical model to perform a recommendation value algorithm to obtain training samples of a second training set; Converting the production result product indicators of the training samples of the second training set into a multi-objective optimization index; If the multi-objective optimization index is greater than a preset value, optimizing the recommendation value algorithm of the first mathematical model using the second training set; If the multi-objective optimization index is less than the preset value, generating a third training set based on the training samples of the second training set and the production scheduling algorithm; generating a second mathematical model based on the third training set; in, Generating the third training set based on the training samples of the second training set and the production scheduling algorithm includes: obtaining the number of training samples in the third training set; if the number of training samples in the third training set is less than a preset value, generating the third training set using the production scheduling algorithm; Wherein, if the number of training samples in the third training set is less than a preset value, generating the third training set by using a production scheduling algorithm includes: If the number of training samples in the third training set is less than the preset value, adjusting the process parameters of the training samples in the second training set using a production scheduling algorithm to obtain trained samples; generating the third training set according to the trained samples; The method also includes: generating a second mathematical model through a machine learning method, using the trained second mathematical model as the final use version, and using the generated second mathematical model to perform a recommendation value algorithm to obtain training samples of a fourth training set, and then converting the production result product indicators of the training samples of the fourth training set into a second multi-objective optimization index until the second multi-objective optimization index is greater than a preset value, and setting the recommended value of the process parameter Xn of the training samples of the fourth training set that finally meets the standard as the final recommended value for the abnormal operating condition, and using it under the abnormal operating condition.

2. The method according to claim 1, characterized in that The trained samples include training samples available under abnormal operating conditions and / or training samples unavailable under abnormal operating conditions, wherein generating the third training set according to the trained samples includes: The training samples available under the abnormal working condition are added to the third training set, and the training samples unavailable under the abnormal working condition are deleted from the third training set to obtain the third training set.

3. A process parameter recommendation device for abnormal working conditions, characterized in that: include: a processing unit configured to generate a first mathematical model based on a first training set, wherein the first training set is generated under normal operating conditions using product design parameters based on a design scorecard during the research and development phase; an acquiring unit, configured to acquire abnormal process parameters under abnormal working conditions, and perform a recommendation value algorithm using the first mathematical model to obtain training samples of a second training set; The processing unit is configured to convert the production result product indicators of the training samples of the second training set into a multi-objective optimization index; if the multi-objective optimization index is greater than a preset value, optimize the recommendation value algorithm of the first mathematical model using the second training set; if the multi-objective optimization index is less than the preset value, generate a third training set using a production scheduling algorithm based on the training samples of the second training set; and generate a second mathematical model based on the third training set; in, Generating the third training set based on the training samples of the second training set and the production scheduling algorithm includes: obtaining the number of training samples in the third training set; if the number of training samples in the third training set is less than a preset value, generating the third training set using the production scheduling algorithm; Wherein, if the number of training samples in the third training set is less than a preset value, generating the third training set by using a production scheduling algorithm includes: If the number of training samples in the third training set is less than the preset value, adjusting the process parameters of the training samples in the second training set using a production scheduling algorithm to obtain trained samples; generating the third training set according to the trained samples; The processing unit is also used to: generate a second mathematical model through a machine learning method, use the trained second mathematical model as the final use version, and use the generated second mathematical model to perform a recommendation value algorithm to obtain training samples of a fourth training set, and then convert the production result product indicators of the training samples of the fourth training set into a second multi-objective optimization index until the second multi-objective optimization index is greater than a preset value, and set the recommended value of the process parameter Xn of the training samples of the fourth training set that finally meets the standard as the final recommended value for the abnormal operating condition, and use it under the abnormal operating condition.

4. A process parameter recommendation device for abnormal working conditions, characterized in that: The device includes a processor and a memory, the memory is used to store a computer program, and the processor is used to call the computer program to execute the method according to any one of claims 1-2.

5. A chip, characterized in that: The chip is a chip in a process parameter recommendation device for abnormal working conditions. The chip includes a processor and an input interface and an output interface connected to the processor. The chip also includes a memory. When the computer program in the memory is executed, the method described in any one of claims 1-2 is executed.

6. A computer-readable storage medium, characterized in that Used to store a computer program, which, when running on a computer, causes the computer to execute the method according to any one of claims 1 to 2.

7. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is run on a computer, the computer is caused to perform the method according to any one of claims 1 to 2.

Images