Flexible job-shop active fault-tolerant scheduling method based on multi-view transfer learning

Abstract

The invention discloses an active fault-tolerant scheduling method for a flexible job shop based on multi-view transfer learning, and the method comprises the steps: collecting time series data of all equipment in the shop in real time, and carrying out the preprocessing for data cleaning and standardization; generating an industrial data interpolation model of an adversarial network through a weighted condition based on gradient penalty, and predicting and filling missing or abnormal equipment state data to complete acquired data; the model is used for carrying out data prediction on the state of a future system; identifying potential problems by using a task migration module based on unscented transformation in combination with a fault analysis technology; a state calculation module is used for calculating the specific state of the intelligent agent in the environment state space according to the complete time sequence data after interpolation; and establishing a flexible job shop dynamic scheduling model, and performing action selection according to the state characteristics of the current production environment based on a multi-view enhanced deep reinforcement learning scheduling module to generate a scheduling scheme. According to the invention, the scheduling efficiency and the fault-tolerant capability in the flexible production environment can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent factories, and in particular relates to an active fault-tolerant scheduling method for a flexible job shop based on multi-perspective transfer learning. Background Art

[0002] With the continuous advancement of advanced manufacturing technology, the manufacturing industry has gradually transformed into service-oriented manufacturing, and workshop production tasks have shown the characteristics of small batches and customization. In this context, traditional large-volume orders have gradually decreased, and the arrival time and quantity of personalized orders have become more random and difficult to predict. In this dynamic environment where the number and time of production orders are difficult to predict, the focus of enterprises has also changed accordingly, from simply focusing on the overall completion time to more complex production management goals, such as ensuring that products are delivered on time and improving the comprehensive utilization efficiency of factory equipment. With the increase in demand for personalized and customized products, companies need a more flexible and responsive scheduling system. Therefore, how to comprehensively consider multiple production goals and formulate scheduling strategies in a dynamic environment has become one of the key issues in the field of smart factories.

[0003] In dynamic workshop scheduling, data loss is a common challenge, especially when sensors fail to provide data in time due to failure or other reasons, the decision of the scheduling system is affected. In smart factories, the uncertainty and dynamic changes of the production environment bring severe challenges to scheduling methods. The scheduling system not only needs to collect and process various production data in real time, but also must continuously collect key information such as equipment status and task progress, so as to adjust the scheduling strategy in real time according to changes in the production environment; especially when production conditions or task orders change, the scheduling method needs to be highly flexible and able to respond quickly to these changes, so as to avoid efficiency decline or even shutdown caused by untimely adjustments. In the dynamic scheduling process, the scheduling system needs to find an appropriate balance between response speed and decision accuracy; relying solely on real-time data for decision-making, when sensors fail or data is missing, will cause the system to be unable to accurately perceive the equipment status, thus affecting the timely response to the rapid changes in the production environment. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a multi-perspective transfer learning flexible job shop active fault-tolerant scheduling method to improve the scheduling efficiency and fault tolerance in the flexible production environment. Through multi-perspective feature extraction, transfer learning and deep reinforcement learning and other technologies, the scheduling process is optimized to cope with downtime caused by equipment failure, incomplete information caused by data loss and dynamic changes in production demand.

[0005] To achieve the above object, the technical solution adopted by the present invention is: a flexible job shop active fault-tolerant scheduling method based on multi-perspective transfer learning, comprising the steps of:

[0006] S10, collects time series data of each device in the workshop in real time, and uses data preprocessing technology to clean and standardize the data;

[0007] S20, through the weighted conditional generation adversarial network industrial data interpolation model based on gradient penalty, the missing or abnormal equipment status data is predicted and filled to complete the collected data; and the model is used to predict the status of the future system;

[0008] S30 uses the task migration module based on traceless transformation and fault analysis technology to identify potential problems; uses the state calculation module to calculate the specific state of the agent in the environment state space based on the complete time series data after interpolation, providing a basis for scheduling decisions;

[0009] S40, establish a dynamic scheduling model for a flexible job shop, and on this basis, based on the multi-perspective enhanced deep reinforcement learning scheduling module, select actions according to the state characteristics of the current production environment and generate a scheduling plan.

[0010] Furthermore, the industrial data interpolation model based on gradient-penalized weighted conditional generative adversarial network adopts a variational autoencoder structure as the generator, establishes a binary classification network based on a multi-layer perceptron as the discriminator, and adopts gradient-penalized weighted conditional generative adversarial network as the training framework of the entire model, and optimizes the parameters of the generator and discriminator through Wasserstein distance.

[0011] Industrial data interpolation model based on gradient penalty weighted conditional generative adversarial network, including:

[0012] S21, encodes the production task type or equipment type data and converts it into a prompt vector in binary format; these prompt vectors are mapped through the pre-coding parameter matrix to obtain a parameter expression suitable for model processing, laying the foundation for subsequent feature extraction.

[0013] S22, multi-view feature extraction: Based on the encoded data, a multi-view feature extraction module is used to enhance the data representation capability;

[0014] S23, spatiotemporal feature encoding: After feature extraction, the temporal dependency and spatial correlation of the data are further captured through the spatiotemporal feature encoding module;

[0015] S24, based on the spatiotemporal encoded features, uses the Transformer decoder to generate interpolation data and performs nonlinear transformation on the features to match the original data dimensions to complete missing value completion. The decoding process includes predictive interpolation and data recovery interpolation, the former is used to complete future data, and the latter is used to complete historical data.

[0016] Furthermore, when using the multi-view feature extraction module to enhance the data representation capability: construct a random mask layer with multiple mask modes to simulate different data missing situations, extract features from the masked data through a feature representor of a long short-term memory network and a linear layer stack, and further use a linear attention mechanism and a multi-layer perceptron to aggregate features, thereby forming a multi-view feature representation.

[0017] Furthermore, the temporal dependency and spatial correlation of the data are captured through the spatiotemporal feature encoding module: in the temporal dimension, the stacked long short-term memory network is used to process the sequence data to model the temporal changes of the data; in the spatial dimension, the convolution and Transformers structure are used to extract features of the data; finally, the temporal features and spatial features of the input sequence are concatenated and fused in a given dimension to achieve joint encoding of multi-dimensional information, providing a comprehensive representation of spatiotemporal information for the interpolation model.

[0018] Furthermore, the Transformer decoder is used to generate interpolation data. The decoding process includes predictive interpolation and data recovery interpolation. The former is used to complete future data, and the latter is used to complete historical data.

[0019] Furthermore, in step S30, the core of the task migration module is to utilize the time series features learned in the data interpolation model and apply them to the fault diagnosis task; then, by introducing the traceless transformation technology, the potential features in the source task are transferred to the target task to achieve adaptation to the fault analysis task;

[0020] S31, feature migration based on untraceable transformation: the spatiotemporally encoded features are used as source task feature inputs, and a set of sigma points are generated through untraceable transformation, which can capture the changes of nonlinear features in different tasks; the generation of sigma points is based on the potential feature distribution of the interpolation task, including generating a central sigma point at the mean position, and generating positive and negative offset sigma points according to the covariance matrix;

[0021] S32, fault identification: using the generated sigma point set, the task migration module transfers the interpolation features to the fault diagnosis decoder to perform fault identification and output the identification results.

[0022] Furthermore, in step S40, a flexible job shop dynamic scheduling model is established, including:

[0023] Basic assumptions: Assume that each device can only process one process at any time, and all workpieces must be completed in a predetermined order; each process is performed on a specific subset of devices, and the processing time is fixed;

[0024] Constraint equations: Based on the basic assumptions, they are mathematically transformed into the constraint equations of the model. First, the equipment utilization rate is used as an indicator to measure the probability of failure, and the maintenance time constraint after failure is set for each device. At the same time, the module replacement time between equipment or workpieces is set as a fixed constant and included in the processing time constraint. The urgency parameter of the workpiece is set, and the model gives priority to high-urgency workpieces during the optimization process.

[0025] Multi-objective optimization: Under the premise of satisfying the above constraint equations, multi-objective optimization is used to achieve the optimal scheduling effect; the optimization goals are: minimizing the total completion time and delay time, and maximizing equipment utilization.

[0026] Furthermore, in step S40, the multi-view enhanced deep reinforcement learning scheduling module includes:

[0027] S41, Modeling: Convert the flexible job shop dynamic scheduling model into a Markov decision process and set the state space, action space and reward function;

[0028] S42, state feature extraction: In order to capture the multi-dimensional state features in the production environment, the state space is constructed by a multi-view feature enhancement module; the multi-view feature enhancement module takes the information in the state space established in step S41 as input, extracts feature vectors using a linear layer, and captures the semantic dependencies between these features through a Transformer network; after feature enhancement, a multi-view state representation is generated;

[0029] S43, Adversarial deep Q network model decision process: The above multi-perspective state representation is directly used as the input of the adversarial deep Q network model. The model calculates the Q value to evaluate the pros and cons of different actions by evaluating the value of the current state and the advantages of the action.

[0030] Furthermore, the reward function includes:

[0031] Input: Average delay time at the current time Maximum delay time Equipment utilization And the total construction period With the next moment

[0032] Process: 1. Judgment and Is it true? If so, mark the delay optimization; 2. Judgment Is it true? If so, mark the equipment utilization rate as optimized; 3. Judge, Is it true? If so, mark the total duration as optimized; 4. According to the status of each optimization flag, accumulate the corresponding reward value; if all optimization flags are True, add an additional reward; if all optimization flags are False, impose a penalty;

[0033] Output: The final calculated reward value.

[0034] Furthermore, the method further includes step S50: introducing a differential privacy mechanism in the online parameter optimization process of the scheduling strategy, and ensuring the model performance while enhancing the data privacy protection capability through differential privacy noise perturbation and differential privacy stochastic gradient descent strategy;

[0035] Differential privacy optimization of online scheduling strategies, including:

[0036] Differential privacy stochastic gradient descent: In parameter optimization, the gradient is clipped and Gaussian noise is added to ensure that the differential privacy requirements are met each time the model is updated;

[0037] Privacy perturbation of input data: In the forward propagation stage, Gaussian noise is added to the input features to control the global sensitivity of the data and enhance privacy protection.

[0038] The beneficial effects of adopting this technical solution are:

[0039] The present invention proposes an active fault-tolerant scheduling method for a flexible job shop based on multi-perspective transfer learning. First, the data interpolation technology and fault analysis method are used to effectively solve the problems of data missing and equipment failure in the production process, thereby improving the reliability and accuracy of the scheduling system. Secondly, the multi-perspective deep reinforcement learning model is used for scheduling optimization to achieve the rational allocation of production resources and improve production efficiency. In addition, a privacy protection mechanism is introduced. By introducing differential privacy noise perturbation and differential privacy stochastic gradient descent strategy in parameter updating, the model's ability in data privacy protection is enhanced while ensuring model performance. Finally, the method shows strong adaptability in a dynamic production environment, can reduce the risk of production interruption caused by equipment failure, and improve equipment utilization. These advantages enable the present invention to effectively improve the overall efficiency and stability of the system when dealing with complex scheduling tasks in flexible job shops.

[0040] The present invention can improve the integrity and accuracy of data processing: a data interpolation model based on a weighted conditional generative adversarial network with gradient penalty is used to fill in the missing parts in the production data. At the same time, through multi-perspective feature extraction, the high-dimensional representation ability of the data is enhanced, providing more accurate data support for subsequent scheduling optimization. The present invention proposes an industrial data interpolation model based on self-supervised learning, which adopts a weighted conditional generative adversarial network training framework with gradient penalty, and combines a multi-perspective embedding module to enrich the feature representation of self-supervised learning. Finally, a multi-task learning mechanism with dynamic weights is introduced, thereby improving the stability of the model in processing multi-variable industrial missing data interpolation tasks. The interpolation model of the present invention has a high degree of responsiveness and can quickly generate reasonable data completion and prediction of future data when data missing occurs; in addition, it can cope with the frequent changes in task orders in the workshop, which puts higher requirements on the robustness of the interpolation model; the interpolation method can fully analyze historical data and combine the information of the current production environment to accurately predict missing data, thereby adapting to the frequent changes in production tasks.

[0041] The present invention designs a working condition and equipment mode coding mechanism, which encodes the working condition information and equipment operation mode into a preset matrix and introduces them into the model as prompt conditions, thereby enhancing the model's adaptability to data distribution under different working conditions.

[0042] The present invention can realize knowledge transfer and reuse: using the task migration module, the learned feature representation is shared between different tasks, and knowledge migration between tasks such as data interpolation and fault diagnosis is realized, thereby enhancing the adaptability of the model in the actual industrial environment. Based on the original model, the present invention introduces a task migration module based on traceless transformation. The traceless transformation module realizes the migration from data interpolation to time series feature analysis tasks by generating Sigma points, promotes information sharing and reuse between different tasks, improves the adaptability and reuse rate of the model, reduces the manpower and time investment required for enterprises to develop new models for different tasks, and reduces costs.

[0043] The present invention can optimize the production scheduling strategy: based on the multi-perspective deep reinforcement learning scheduling module, the production resources and tasks are optimally allocated to ensure that the production efficiency can be maximized when the production environment changes such as equipment failure. This module combines the adversarial deep Q network and the multi-perspective Transformers feature enhancement module to improve the ability to understand the production status, thereby making better scheduling decisions. The present invention proposes a representation learning module based on multi-perspective Transformers, which improves the decision-making ability of the scheduling algorithm in complex dynamic environments by multi-level modeling of state space features and capturing global dependencies, and inputs them as high-level states into DQN, thereby ensuring that it can still maintain efficient and stable operation in a complex and changing production environment.

[0044] The present invention introduces a privacy protection mechanism. By introducing differential privacy noise perturbation and differential privacy stochastic gradient descent strategy in parameter updating, the model's ability in data privacy protection is enhanced while ensuring model performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A schematic diagram of the framework principle of a flexible job shop active fault-tolerant scheduling method based on multi-perspective transfer learning of the present invention;

[0046] Figure 2 Schematic diagram of a data recovery interpolation and predictive interpolation model in an embodiment of the present invention;

[0047] Figure 3 Schematic diagram of a multi-view missing data feature embedding module in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described below with reference to the accompanying drawings.

[0049] In this embodiment, see Figure 1 As shown, the present invention proposes a flexible job shop active fault-tolerant scheduling method based on multi-perspective transfer learning, comprising the steps of:

[0050] S10, collects time series data of each device in the workshop in real time, and uses data preprocessing technology to clean and standardize the data;

[0051] S20, through the weighted conditional generation adversarial network industrial data interpolation model based on gradient penalty, the missing or abnormal equipment status data is predicted and filled to complete the collected data; and the model is used to predict the status of the future system;

[0052] S30 uses the task migration module based on traceless transformation and fault analysis technology to identify potential problems; uses the state calculation module to calculate the specific state of the agent in the environment state space based on the complete time series data after interpolation, providing a basis for scheduling decisions;

[0053] S40, establish a dynamic scheduling model for a flexible job shop, and on this basis, based on the multi-perspective enhanced deep reinforcement learning scheduling module, select actions according to the state characteristics of the current production environment and generate a scheduling plan.

[0054] As an optimization scheme for the above embodiment, an industrial data interpolation model based on a weighted conditional generative adversarial network with gradient penalty adopts a variational autoencoder (VAE) as a generator, establishes a binary classification network based on a multilayer perceptron (MLP) as a discriminator, and adopts a weighted conditional generative adversarial network with gradient penalty as the training framework of the entire model, and optimizes the parameters of the generator and the discriminator through the Wasserstein distance.

[0055] Specifically, the industrial data interpolation model based on the weighted conditional generative adversarial network with gradient penalty is as follows: Figure 2 As shown, including:

[0056] S21, encodes the production task type or equipment type data and converts it into a prompt vector in binary format; these prompt vectors are mapped through the pre-coding parameter matrix to obtain a parameter expression suitable for model processing, laying the foundation for subsequent feature extraction.

[0057] First, define the same data domain in the training dataset Different production processes or equipment types, denoted by For each class i , which is converted into Binary Vector Defined as:

[0058]

[0059] in, Representation vector The value of the jth dimension. This ensures that each category has a unique representation in the encoding vector, avoiding the impact of potential order relationships between categories on model training.

[0060] is the one-hot encoded prompt vector The present invention introduces the precoding parameter matrix The mapping process can be described as:

[0061]

[0062] in It is the parameter expression vector after mapping, which is used for subsequent model processing.

[0063] Then, the input data is normalized to form a standardized data set using the Z normalization method, as shown in the following formula:

[0064]

[0065] Among them, μimpu is the mean of the input data, σ impu is the variance of the input data. Through this industrial preprocessing method, the unknown industrial input data conforms to the assumption of normal distribution and meets the prerequisite of VAE. By precoding the production tasks or equipment types and preprocessing the input data, it helps the model to obtain more representative features in the samples during the training process, thereby improving the performance of the model in processing interpolated data in the context of personalized customization.

[0066] S22, multi-view feature extraction: Based on the encoded data, a multi-view feature extraction module is used to enhance the data representation capability.

[0067] Specifically, when using the multi-view feature extraction module to enhance the data representation capability: construct a random mask layer with multiple mask modes to simulate different data missing situations, extract features from the masked data through a feature representor stacked with LSTM and linear layers, and further use the linear attention mechanism and MLP for feature aggregation to form a multi-view feature representation.

[0068] In order to enhance the generalization ability of the model under limited training samples and thus improve the accuracy of data interpolation, the present invention proposes a multi-view feature extraction module to enrich the data representation ability. Several different mask patterns are constructed through random mask layers to simulate various data missing situations. The time window size is Input sample Through random mask operation Mask impu , obtain a set of industrial mask data under different mask conditions,

[0069]

[0070] This will allow us to obtain more observational data samples that retain the original temporal structure and pattern characteristics.

[0071] Then, the masked industrial data is input into Figure 3 The multi-view feature extraction module shown in FIG. 1 is composed of several feature representation heads and feature aggregation heads. The module is composed of several feature representors and feature aggregators. Each feature representor adopts the design of long short-term memory network (LSTM) and linear layer stacking to process input data under different mask conditions and extract its feature representation, and obtain the fused feature representation vector through cat operation. The process can be expressed as:

[0072]

[0073] in, represents the i-th feature extractor. The feature aggregator consists of a linear attention layer and an MLP

[0074] The concatenated feature representations are weighted and aggregated through a linear attention mechanism to obtain a weighted feature vector.

[0075] Subsequently, the aggregated feature vector is transformed nonlinearly through MLP to achieve dimensionality control. This process can be expressed as:

[0076]

[0077] The motivation for proposing this strategy in this paper is that in practical applications, the migration fine-tuning method of adding a linear layer to the original model to perform feature dimension shaping often reduces the model's expressiveness and computational efficiency. By adopting a multi-view feature extraction module, these problems can be effectively avoided while enhancing the richness and diversity of feature representation.

[0078] S23, spatiotemporal feature encoding: After feature extraction, the temporal dependency and spatial correlation of the data are further captured through the spatiotemporal feature encoding module.

[0079] Specifically, the temporal dependency and spatial correlation of the data are captured through the spatiotemporal feature encoding module: in the temporal dimension, stacked LSTM is used to process the sequence data to model the temporal changes of the data; in the spatial dimension, convolution and Transformers structures are used to extract features of the data; finally, the temporal features and spatial features of the input sequence are concatenated and fused in a given dimension to achieve joint encoding of multi-dimensional information, providing a comprehensive representation of spatiotemporal information for the interpolation model.

[0080] In order to capture the temporal structure dependency and explore the spatial correlation between the measurement data of the same device, the present invention designs a spatiotemporal feature extraction module. The module uses a stacked LSTM to capture the temporal dependency. At the same time, a convolution-Transformers structure is used to aggregate spatial features. For the time dimension, a stacked LSTM structure is used to process the input sequence data to generate a temporal feature representation. As for the spatial dimension, the present invention first uses convolution operation to capture the local dependencies in space, and then further processes the spatial features through Transformers. The process can be described as follows:

[0081]

[0082] in, represents the convolution kernel, * represents the convolution operation, represents the SiLU activation function, Represents the convolution bias. The specific form of the convolution operation can be expressed as:

[0083]

[0084] Among them, M, N are the height and width of the convolution kernel, i, j are the spatial position index of the feature map. After processing by the spatiotemporal feature encoder, we can get:

[0085]

[0086] in, represents high-dimensional time features, Represents high-dimensional space features, Represents the high-dimensional spatiotemporal features obtained through the Cat operation. Then, the multi-head attention layer and MLP are used to obtain The mean of the low-dimensional representation of and variance

[0087] S24, based on the spatiotemporal encoded features, uses the Transformer decoder to generate interpolation data and performs nonlinear transformation on the features to match the original data dimensions to complete missing value completion. The decoding process includes predictive interpolation and data recovery interpolation, the former is used to complete future data, and the latter is used to complete historical data.

[0088] Specifically, a Transformer decoder is used to generate interpolation data. The decoding process includes predictive interpolation and data recovery interpolation. The former is used to complete future data, and the latter is used to complete historical data.

[0089] The present invention adopts a Transformer-based decoder structure and combines it with MLP to build a decoding network. Specifically, the Transformer decoder first processes low-dimensional features through a multi-head attention mechanism to capture complex spatiotemporal dependencies, and then converts the processed features into interpolated values ​​that match the dimensions of the original data through MLP. The process can be expressed as:

[0090]

[0091] The advantage of this design is that the dimensions of the MLP can be dynamically adjusted according to the different downstream tasks to adapt to different task requirements, thereby improving the flexibility and generalization ability of the model. In addition, this design also enhances the model's ability to handle complex spatiotemporal dependencies, which helps improve interpolation accuracy.

[0092] In the training process of the industrial data interpolation model of the gradient-penalized weighted conditional generative adversarial network: the gradient-penalized weighted conditional generative adversarial network is used as the training framework of the generator, aiming to optimize the adversarial training of the generator and the discriminator, and solve the gradient vanishing problem that may occur in the generator during the training process. On the basis of the traditional WCGAN, the gradient penalty technology is introduced to further enhance the stability and training effect of the model. During the entire training process, the present invention generates reconstructed labels through random masks and combines the construction mechanism of predicted labels to realize self-supervised learning based on time series data. The advantage of self-supervised learning is that no additional manual annotation is required, and training samples can be generated only through some missing information in the original data, which reduces the dependence on large-scale labeled data, and can effectively improve the generalization ability of the model and the accuracy of data interpolation.

[0093] The core of the training strategy of the present invention is to use VAE as a generator while introducing a weighted conditional generative adversarial network with gradient penalty, and enhance the generation ability of the generator and the discrimination ability of the discriminator by introducing a gradient penalty strategy. Under this framework, VAE, as a generator, can more effectively learn the potential distribution of time series data and ensure that the generated data has reasonable time series consistency and accuracy. VAE, as an extended model of AE, introduces a probability generation mechanism and regularization constraints of latent variables on the basis of traditional AE, thereby improving the model's modeling ability of data distribution. The latent encoding of VAE not only contains the mean and variance, but also ensures the structured distribution of the latent space through regularization constraints, providing favorable conditions for knowledge transfer. The distribution of the latent variable z is regularized by KL divergence to make it close to the prior distribution, thereby making knowledge transfer between different tasks or equipment types possible. In this study, by sharing the structure of the latent space, knowledge transfer between industrial time series data and tasks can be achieved, and the model knowledge reuse rate is improved.

[0094] However, VAE faces problems such as weak temporal consistency of generated data and inaccurate modeling of temporal dynamics in industrial time series data interpolation tasks. To solve this problem, the present invention introduces an adversarial generation mechanism. Through WCGAN, the generator can minimize the distribution of real data p. data and generate data distribution p g While ensuring the accuracy of data interpolation, WCGAN can provide more stable gradients than traditional GANs, avoid mode collapse and gradient instability problems, and thus train the generator more effectively.

[0095] As an optimization solution of the above embodiment, in step S30, the core of the task migration module is to use the time series features learned in the data interpolation model and apply them to the fault diagnosis task; then, by introducing the traceless transformation technology, the potential features in the source task are transferred to the target task to achieve adaptation to the fault analysis task;

[0096] S31, feature migration based on untraceable transformation: the spatiotemporally encoded features are used as source task feature inputs, and a set of sigma points are generated through untraceable transformation, which can capture the changes of nonlinear features in different tasks; the generation of sigma points is based on the potential feature distribution of the interpolation task, including generating a central sigma point at the mean position, and generating positive and negative offset sigma points according to the covariance matrix;

[0097] S32, fault identification: using the generated sigma point set, the task migration module transfers the interpolation features to the fault diagnosis decoder to perform fault identification and output the identification results.

[0098] In task transfer, the present invention focuses on how to share knowledge between different but related tasks. Specifically, although the source task and the target task have different target output spaces, they share similar input feature spaces and feature representations. With target tasks in and denote the input space of the source task and the target task respectively, Respectively represent the corresponding output space. Generally speaking, in the task context of the present invention, the input space is the same, but from the perspective of rigor, it is assumed that the two are similar, that is, So they can share the same feature encoder The source task model trained in the data interpolation task can be expressed as:

[0099]

[0100] In order to transfer the knowledge learned by the original encoder to the fault diagnosis task, the present invention adds a new decoder for the target task. However, considering that the potential space distribution of different tasks may be different, it is not advisable to directly use the latent variables generated by the encoder. It is impossible to achieve the desired effect in the fault diagnosis task. In order to solve this problem, the present invention introduces unscented transformation as the core idea of ​​unscented Kalman filtering. Unscented transformation can more accurately estimate the statistical characteristics of random variables after nonlinear transformation by generating a set of sigma points without calculating high-order derivatives. For n-dimensional independent and identically distributed latent variables Its overall mean is The covariance matrix is Then the steps to generate the sigma point are:

[0101] 1. Generate a sigma point at the mean position

[0102] 2. Generate 2n offset points based on the covariance matrix, of which n are positive offset points and the other n are negative offset points. These offset points are generated by transforming the square root of the covariance matrix, as shown below:

[0103]

[0104] in represents the square root of the covariance matrix, λ it Indicates that the scaling parameter is used to control the distribution range of the sigma points.

[0105] Through the above steps, the present invention obtains a richer data representation to capture the potential feature changes between the source domain and the target domain. Input to the new decoder In the task, the fault diagnosis results of the target domain are obtained. At this time, the loss function of task migration can be expressed as:

[0106]

[0107] in, Represents the loss function of the fault diagnosis task. Through this migration module, the time series feature knowledge learned in the data interpolation task is reused, which enhances the adaptability of the model between different tasks in the same domain (such as from data interpolation tasks to fault diagnosis tasks), thereby improving the application value of the model in actual industrial environments.

[0108] As an optimization solution of the above embodiment, in step S40, a flexible job shop dynamic scheduling model is established, including:

[0109] Basic assumptions: Assume that each device can only process one process at any time, and all workpieces must be completed in a predetermined order; each process is performed on a specific subset of devices, and the processing time is fixed;

[0110] Constraint equations: Based on the basic assumptions, they are mathematically transformed into the constraint equations of the model. First, the equipment utilization rate is used as an indicator to measure the probability of failure. The post-failure maintenance time constraint is set for each device to ensure that the equipment can resume normal operation according to the scheduled maintenance time after a failure, thereby reducing the impact of the failure on scheduling. At the same time, the module replacement time between equipment or workpieces is set as a fixed constant and included in the processing time constraint to improve scheduling accuracy and stability. The urgency parameter of the workpiece is set, and the model gives priority to high-urgency workpieces during the optimization process, thereby improving the scheduling response capability to emergency tasks.

[0111] Multi-objective optimization: Under the premise of satisfying the above constraint equations, multi-objective optimization is used to achieve the optimal scheduling effect; the optimization goals are: minimizing the total completion time and delay time, and maximizing equipment utilization.

[0112] In step S40, the multi-view enhanced deep reinforcement learning scheduling module includes:

[0113] S41, Modeling: Convert the flexible job shop dynamic scheduling model into a Markov decision process and set the state space, action space and reward function;

[0114] First, the state space comprehensively models the dynamic state of each device and operation in the system, including multi-dimensional information such as equipment availability, operation progress, and process requirements; secondly, the action space is defined as the scheduling decision rules that the system can take under a given state to ensure the smooth completion of the scheduling task. The reward function combines the requirements of multi-objective optimization and comprehensively considers factors such as production cycle, equipment utilization, and operation delay to ensure the adaptability of the model to multi-objective scheduling tasks.

[0115] (1) State space:

[0116] The state space is the core element for the intelligent agent to perceive the workshop production environment, and it is also the basis for the intelligent agent to make correct scheduling decisions. Therefore, when designing the state space, the present invention follows the principle of combining simplicity and effectiveness, and strives to use the least state features to describe the production process of the dynamic flexible workshop, so as to reduce the complexity of the model while meeting the scheduling objectives. As shown in Table 1, the present invention selects 12 important features as state features.

[0117] Table 2 State space description

[0118]

[0119] (2) Action Space

[0120] The present invention designs seven different scheduling rules for multi-objective optimization problems, each of which prioritizes jobs based on different criteria. In this way, prior knowledge is integrated into the action space design, thereby simplifying the complexity of action selection. It should be pointed out that an overly redundant action space will increase the complexity of learning and the computational burden. The designed action space is shown in Table 2.

[0121] Table 2 Action space description

[0122]

[0123]

[0124] (3) Reward Function

[0125] In order to guide the intelligent agent to make the best scheduling decision in a dynamic flexible job shop environment, this paper designs a multi-objective optimization reward function based on multi-objective reinforcement learning theory. The reward function provides a structured reward mechanism by evaluating delays, equipment utilization, and total construction period. Its design idea is derived from the Pareto optimality theory, which seeks a balance between multiple conflicting objectives and encourages the intelligent agent to explore the optimal scheduling strategy.

[0126] The reward function includes:

[0127] Input: Average delay time at the current time Maximum delay time Equipment utilization And the total construction period With the next moment

[0128] Process: 1. Judgment and Is it true? If so, mark the delay optimization; 2. Judgment Is it true? If so, mark the equipment utilization rate as optimized; 3. Judge, Is it true? If so, mark the total construction period as optimized; 4. According to the status of each optimization flag, accumulate the corresponding reward value; if all optimization flags are True, add an additional reward; if all optimization flags are False, impose a penalty;

[0129] Output: The final calculated reward value.

[0130] S42, state feature extraction: In order to capture the multi-dimensional state features in the production environment, the state space is constructed by a multi-view feature enhancement module; the multi-view feature enhancement module takes the information in the state space established in step S41 as input, extracts feature vectors using a linear layer, and captures the semantic dependencies between these features through a Transformer network; after feature enhancement, a multi-view state representation is generated;

[0131] S43, Dueling DQN model decision process: The above multi-perspective state representation is directly used as the input of the Dueling DQN model. The model calculates the Q value to evaluate the pros and cons of different actions by evaluating the value of the current state and the advantage of the action.

[0132] As an optimization scheme of the above embodiment, in the parameter optimization process of deep reinforcement learning, the present invention also introduces a privacy protection mechanism, combined with differential privacy technology, to ensure the security protection of production feature data during the training process, thereby improving the security of the model in actual industrial scenarios.

[0133] The method further includes step S50: introducing a differential privacy mechanism in the online parameter optimization process of the scheduling strategy, and ensuring the model performance while enhancing the data privacy protection capability through differential privacy noise perturbation and differential privacy stochastic gradient descent strategy;

[0134] Differential privacy optimization of online scheduling strategies, including:

[0135] Differential privacy stochastic gradient descent: In parameter optimization, the gradient is clipped and Gaussian noise is added to ensure that the differential privacy requirements are met each time the model is updated;

[0136] Privacy perturbation of input data: In the forward propagation stage, Gaussian noise is added to the input features to control the global sensitivity of the data and enhance privacy protection.

[0137] The present invention introduces a differential privacy mechanism to protect sensitive information in the scheduling process from being leaked during the training process. In a dynamic workshop scheduling environment, the scheduling tasks and environmental states are highly dynamic. Therefore, any two scheduling tasks can be defined as adjacent training sets. The differential privacy mechanism ensures that when the model faces these adjacent data sets, its output and parameter updates will not be obvious, thereby protecting the sensitive information of a single scheduling task. The differential privacy mechanism introduced in this method is implemented through two methods. First, in each iteration of model training, a differential privacy stochastic gradient descent algorithm is used to clip the gradient and add noise. For the gradient of parameter θ Its clipped gradient can be expressed as:

[0138]

[0139] in, is the gradient clipping threshold. Gaussian noise is added to the clipped gradient to meet the differential privacy requirements. The process is expressed as:

[0140]

[0141] in, Represents the noise intensity parameter. The gradient update expression after adding noise is:

[0142]

[0143] Among them, γ fjsp Represents the learning rate. In order to further protect the privacy of the input data, Gaussian noise is added to the input data during the forward propagation of the model. The process is expressed as:

[0144]

[0145] Among them, δ fjsp ,∈ fjsp represents the differential privacy parameter, η fisp Represents the global sensitivity of the input data.

[0146] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A flexible job shop active fault-tolerant scheduling method based on multi-perspective transfer learning, characterized in that: Includes steps: S10, collects time series data of each device in the workshop in real time, and uses data preprocessing technology to clean and standardize the data; S20, through the weighted conditional generation adversarial network industrial data interpolation model based on gradient penalty, the missing or abnormal equipment status data is predicted and filled to complete the collected data; and the model is used to predict the status of the future system; S30, using the task migration module based on traceless transformation combined with fault analysis technology to identify potential problems; Using the state calculation module, the specific state of the agent in the environment state space is calculated based on the complete time series data after interpolation, providing a basis for scheduling decisions; S40, establish a dynamic scheduling model for a flexible job shop, and on this basis, based on the multi-perspective enhanced deep reinforcement learning scheduling module, select actions according to the state characteristics of the current production environment and generate a scheduling plan.

2. According to the multi-perspective transfer learning flexible job shop active fault-tolerant scheduling method of claim 1, it is characterized in that: The industrial data interpolation model based on gradient-penalized weighted conditional generative adversarial network uses a variational autoencoder as a generator, establishes a binary classification network based on a multi-layer perceptron as a discriminator, uses a gradient-penalized weighted conditional generative adversarial network as the training framework of the entire model, and optimizes the parameters of the generator and discriminator through the Wasserstein distance; Industrial data interpolation model based on gradient penalty weighted conditional generative adversarial network, including: S21, encoding the production task type or equipment type data and converting it into a prompt vector in binary format; These prompt vectors are mapped through the precoding parameter matrix to obtain parameter expressions suitable for model processing; S22, multi-view feature extraction: Based on the encoded data, a multi-view feature extraction module is used to enhance the data representation capability; S23, spatiotemporal feature encoding: After feature extraction, the temporal dependency and spatial correlation of the data are further captured through the spatiotemporal feature encoding module; S24, based on the spatiotemporal encoded features, uses the Transformer decoder to generate interpolation data and performs nonlinear transformation on the features to match the original data dimensions to complete missing value completion. The decoding process includes predictive interpolation and data recovery interpolation, the former is used to complete future data, and the latter is used to complete historical data.

3. According to the multi-perspective transfer learning flexible job shop active fault-tolerant scheduling method of claim 2, it is characterized in that: When using the multi-view feature extraction module to enhance data representation capabilities: construct random mask layers with multiple mask modes to simulate different data missing situations, extract features from the masked data through a feature representor that uses a long short-term memory network and a linear layer stack, and further use a linear attention mechanism and a multi-layer perceptron to aggregate features to form a multi-view feature representation.

4. According to the multi-perspective transfer learning flexible job shop active fault-tolerant scheduling method of claim 2, it is characterized in that: The temporal dependency and spatial correlation of data are captured through the spatiotemporal feature encoding module: in the temporal dimension, stacked long short-term memory networks are used to process sequence data to model the temporal changes of data; in the spatial dimension, convolution and Transformers structures are used to extract features of the data; finally, the input sequence is concatenated and fused with temporal features and spatial features in a given dimension to achieve joint encoding of multi-dimensional information, providing a comprehensive representation of spatiotemporal information for the interpolation model.

5. The method for active fault-tolerant scheduling of flexible job shops based on multi-perspective transfer learning according to claim 2 is characterized in that: The Transformer decoder is used to generate interpolation data. The decoding process includes predictive interpolation and data recovery interpolation. The former is used to complete future data, and the latter is used to complete historical data.

6. The method for active fault-tolerant scheduling of flexible job shops based on multi-perspective transfer learning according to claim 1 is characterized in that: In step S30, the core of the task migration module is to use the time series features learned in the data interpolation model and apply them to the fault diagnosis task; then, by introducing the traceless transformation technology, the potential features in the source task are transferred to the target task to achieve adaptation to the fault analysis task; S31, feature migration based on untraceable transformation: the spatiotemporally encoded features are used as source task feature inputs, and a set of sigma points are generated through untraceable transformation, which can capture the changes of nonlinear features in different tasks; the generation of sigma points is based on the potential feature distribution of the interpolation task, including generating a central sigma point at the mean position, and generating positive and negative offset sigma points according to the covariance matrix; S32, fault identification: using the generated sigma point set, the task migration module transfers the interpolation features to the fault diagnosis decoder to perform fault identification and output the identification results.

7. The method for active fault-tolerant scheduling of flexible job shops based on multi-perspective transfer learning according to claim 1 is characterized in that: In step S40, a flexible job shop dynamic scheduling model is established, including: Basic assumptions: Assume that each device can only process one process at any time, and all workpieces must be completed in a predetermined order; each process is performed on a specific subset of devices, and the processing time is fixed; Constraint equations: Based on the basic assumptions, they are mathematically transformed into the constraint equations of the model. First, the equipment utilization rate is used as an indicator to measure the probability of failure, and the maintenance time constraint after failure is set for each device. At the same time, the module replacement time between equipment or workpieces is set as a fixed constant and included in the processing time constraint. The urgency parameter of the workpiece is set, and the model gives priority to high-urgency workpieces during the optimization process. Multi-objective optimization: Under the premise of satisfying the above constraint equations, multi-objective optimization is used to achieve the optimal scheduling effect; the optimization goals are: minimizing the total completion time and delay time, and maximizing equipment utilization.

8. The method for active fault-tolerant scheduling of flexible job shops based on multi-perspective transfer learning according to claim 7 is characterized in that: In step S40, the multi-view enhanced deep reinforcement learning scheduling module includes: S41, Modeling: Convert the flexible job shop dynamic scheduling model into a Markov decision process and set the state space, action space and reward function; S42, state feature extraction: In order to capture the multi-dimensional state features in the production environment, the state space is constructed by a multi-view feature enhancement module; the multi-view feature enhancement module takes the information in the state space established in step S41 as input, extracts feature vectors using a linear layer, and captures the semantic dependencies between these features through a Transformer network; after feature enhancement, a multi-view state representation is generated; S43, Adversarial deep Q network model decision process: The above multi-perspective state representation is directly used as the input of the adversarial deep Q network model. The model calculates the Q value to evaluate the pros and cons of different actions by evaluating the value of the current state and the advantages of the action.

9. The method for active fault-tolerant scheduling of flexible job shops based on multi-perspective transfer learning according to claim 8 is characterized in that: The reward function includes: Input: Average delay time at the current time Maximum delay time Equipment utilization And the total construction period With the next moment Process:

1. Judgment and Is it true? If so, mark the delay optimization; 2. Judgment Is it true? If so, mark the equipment utilization rate as optimized; 3. Judge, Is it true? If so, mark the total duration as optimized; 4. According to the status of each optimization flag, accumulate the corresponding reward value; if all optimization flags are True, add an additional reward; if all optimization flags are False, impose a penalty; Output: The final calculated reward value.

10. The method for active fault-tolerant scheduling of flexible job shops based on multi-perspective transfer learning according to claim 1, characterized in that: The method further includes step S50: introducing a differential privacy mechanism in the online parameter optimization process of the scheduling strategy, and ensuring the model performance while enhancing the data privacy protection capability through differential privacy noise perturbation and differential privacy stochastic gradient descent strategy; Differential privacy optimization of online scheduling strategies, including: Differential privacy stochastic gradient descent: In parameter optimization, the gradient is clipped and Gaussian noise is added to ensure that the differential privacy requirements are met each time the model is updated; Privacy perturbation of input data: In the forward propagation stage, Gaussian noise is added to the input features to control the global sensitivity of the data and enhance privacy protection.

Images