Operation control method and device, electronic equipment and computer program product

Abstract

The invention relates to an operation control method and device, electronic equipment and a computer program product, and relates to the technical field of machine learning. The method comprises the following steps: acquiring operation image data, and performing feature extraction on the operation image data to obtain an operation image feature vector; acquiring historical operation service data and a pre-constructed initial model, wherein the initial model comprises a long-short term memory network structure; training the initial model based on the operation image feature vector and historical operation business data to obtain an operation management and control prediction model; and inputting the obtained real-time operation image data and real-time operation business data into the operation management and control prediction model to obtain operation state data which is used for determining an operation management and control scheme. According to the method, key information in an operation link is accurately extracted through image recognition, and the powerful data analysis capability of the model is combined, so that the enterprise operation state can be more accurately predicted, errors caused by manual judgment are reduced, and the accuracy of decision making is improved.

Description

Operational control methods and devices, electronic equipment and computer program products Technical Field

[0001] This disclosure relates to the field of machine learning technology, and more specifically, to an operation control method, operation control device, electronic device, and computer program product. Background Technology

[0002] As IoT technology matures, enterprises are increasingly demanding data collection, analysis, and real-time monitoring. Simultaneously, image information needs to be analyzed and processed. However, existing recognition technologies are superficial, suffering from errors, complex configurations, and limited functionality, making it difficult to meet diverse application needs.

[0003] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this disclosure is to provide an operation control method, operation control device, electronic device, computer-readable storage medium, and computer program product, thereby overcoming, to at least a certain extent, the problems of low efficiency, poor accuracy, insufficient real-time performance, and inability to cope with a large number of complex business processes that traditional operation control solutions usually rely on human experience and simple data statistical analysis.

[0005] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0006] According to a first aspect of this disclosure, an operation management and control method is provided, comprising: acquiring operation image data; extracting features from the operation image data to obtain operation image feature vectors; acquiring historical operation business data and a pre-constructed initial model, wherein the historical operation business data has temporal characteristics and the initial model includes a long short-term memory network structure; training the initial model based on the operation image feature vectors and the historical operation business data to obtain an operation management and control prediction model; acquiring real-time operation image data and real-time operation business data; inputting the real-time operation image data and real-time operation business data into the operation management and control prediction model to obtain operation status data, wherein the operation status data is used to determine an operation management and control scheme.

[0007] In one exemplary embodiment of this disclosure, the step of acquiring operational image data and extracting features from the operational image data to obtain an operational image feature vector includes: acquiring initial operational image data; performing image preprocessing on the initial operational image data to obtain the operational image data; extracting features from the operational image data to obtain global features and local features corresponding to the operational image data; and generating the operational image feature vector based on the global features and the local features.

[0008] In one exemplary embodiment of this disclosure, the step of obtaining initial operational image data includes: determining the operational area to be collected based on operational management and control requirements, and configuring an image acquisition device in the operational area to be collected; and acquiring operational-related images of the operational area to be collected through the image acquisition device as the initial operational image data.

[0009] In one exemplary embodiment of this disclosure, the step of training the initial model based on the time-series features of the model input data using the Long Short-Term Memory network structure to obtain an operation management prediction model includes: sorting the historical operation business data according to chronological order to obtain operation business sequence data; labeling the operation business sequence data to obtain labeled operation business data; fusing the labeled operation business data with the operation image feature vector to obtain fused input data; determining the operation prediction result corresponding to the fused input data using the initial model; and training the initial model based on a pre-constructed loss function, the operation prediction result, and the historical operation business data until the training termination condition is met to obtain the operation management prediction model.

[0010] In one exemplary embodiment of this disclosure, the method further includes: obtaining the model prediction result corresponding to the operation management and control prediction model, and the operation feedback result in the business operation scenario; determining the model optimization target based on the model prediction result and the operation feedback result; and performing optimization and adjustment operations on the operation management and control prediction model based on the model optimization target to obtain the adjusted operation management and control prediction model, wherein the optimization and adjustment operations include one or more of adjusting the model structure, increasing the amount of training data, and optimizing the model parameters.

[0011] In one exemplary embodiment of this disclosure, the method further includes: obtaining a pre-configured operational status threshold; comparing the operational status data with the operational status threshold to obtain an operational indicator difference; generating operational early warning information when the operational indicator difference is not within the difference deviation range; or obtaining a pre-built operational early warning analysis model; and analyzing the operational status data using the operational early warning analysis model to generate operational early warning information.

[0012] In one exemplary embodiment of this disclosure, the method further includes: comparing and analyzing the operational status data with historical operational business data to obtain operational status change trend information; and generating the operational management and control plan based on the operational status change trend information.

[0013] According to a second aspect of this disclosure, an operation control device is provided, comprising: an image feature extraction module for acquiring operation image data and extracting features from the operation image data to obtain an operation image feature vector; a data model acquisition module for acquiring historical operation business data and a pre-constructed initial model, wherein the historical operation business data has temporal characteristics and the initial model includes a long short-term memory network structure; a model training module for training the initial model based on the operation image feature vector and the historical operation business data to obtain an operation control prediction model; a real-time data acquisition module for acquiring real-time operation image data and real-time operation business data; and an operation status prediction module for inputting the real-time operation image data and real-time operation business data into the operation control prediction model to obtain operation status data, wherein the operation status data is used to determine an operation control scheme.

[0014] In one exemplary embodiment of this disclosure, the image feature extraction module includes an image feature extraction unit, configured to: acquire initial operational image data; perform image preprocessing on the initial operational image data to obtain the operational image data; extract features from the operational image data to obtain global features and local features corresponding to the operational image data; and generate the operational image feature vector based on the global features and the local features.

[0015] In one exemplary embodiment of this disclosure, the image feature extraction unit includes an image acquisition subunit, configured to: determine the operational area to be collected based on operational management requirements, and configure an image acquisition device in the operational area to be collected; acquire operational-related images of the operational area to be collected through the image acquisition device, as the initial operational image data.

[0016] In one exemplary embodiment of this disclosure, the model training module includes a model training unit, configured to: sort the historical operational business data according to chronological order to obtain operational business sequence data; label the operational business sequence data to obtain labeled operational business data; fuse the labeled operational business data with the operational image feature vector to obtain fused input data; determine the operational prediction result corresponding to the fused input data using the initial model; and train the initial model based on a pre-constructed loss function, the operational prediction result, and the historical operational business data until the training termination condition is met to obtain the operational management prediction model.

[0017] In one exemplary embodiment of this disclosure, the operation management and control device further includes a model optimization module, configured to: obtain the model prediction result corresponding to the operation management and control prediction model, and the operation feedback result in the business operation scenario; determine the model optimization target based on the model prediction result and the operation feedback result; and perform optimization and adjustment operations on the operation management and control prediction model based on the model optimization target to obtain an adjusted operation management and control prediction model, wherein the optimization and adjustment operations include one or more of adjusting the model structure, increasing the amount of training data, and optimizing the model parameters.

[0018] In one exemplary embodiment of this disclosure, the operation control device further includes an early warning information generation module, configured to: acquire a pre-configured operation status threshold; compare the operation status data with the operation status threshold to obtain an operation indicator difference; generate operation early warning information when the operation indicator difference is not within the difference deviation range; or acquire a pre-built operation early warning analysis model; and analyze the operation status data using the operation early warning analysis model to generate operation early warning information.

[0019] In one exemplary embodiment of this disclosure, the operation management and control device further includes a management and control scheme generation module, used to: compare and analyze the operation status data with historical operation business data to obtain operation status change trend information; and generate the operation management and control scheme based on the operation status change trend information.

[0020] According to a third aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory storing computer-readable instructions that, when executed by the processor, implement the operation control method according to any one of the preceding claims.

[0021] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the operation control method according to any one of the preceding claims.

[0022] According to a fifth aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the operation control method described in any of the preceding claims.

[0023] The technical solutions provided in this disclosure may include the following beneficial effects: The operation control method in the exemplary embodiments of this disclosure, on the one hand, can accurately extract key information in the operation process through image recognition, and combined with the powerful data analysis capabilities of the model, can more accurately predict the enterprise's operating status, reduce errors in human judgment, and improve the accuracy of decision-making; on the other hand, by collecting and processing image and business data in real time, the model can quickly evaluate the operating situation, enabling relevant personnel to keep abreast of the enterprise's operational dynamics, respond in a timely manner, and avoid the escalation of problems.

[0024] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without creative effort. In the drawings: Figure 1 schematically illustrates a flowchart of an operation control method according to an exemplary embodiment of this disclosure; Figure 2 schematically illustrates a scheme diagram of deploying image acquisition equipment in special areas such as ordinary river sections or cross-sections according to an exemplary embodiment of this disclosure; Figure 3 schematically illustrates a scheme diagram of deploying image acquisition equipment at sand dredger docking points or large river / lake scenes according to an exemplary embodiment of this disclosure; Figure 4 schematically illustrates a scheme diagram of deploying image acquisition equipment in sand pits according to an exemplary embodiment of this disclosure; Figure 5 schematically illustrates a block diagram of an operation control device according to an exemplary embodiment of this disclosure; Figure 6 schematically illustrates a computer-readable storage medium according to an exemplary embodiment of this disclosure; Figure 7 schematically illustrates a block diagram of an electronic device according to an exemplary embodiment of this disclosure. Detailed Implementation

[0026] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0027] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of the specific details described, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known structures, methods, apparatuses, implementations, materials, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0028] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, or in one or more software-hardened modules, or in different network and / or processor devices and / or microcontroller devices.

[0029] In some related solutions, while collecting data, enterprises need to analyze and process image information. However, existing recognition technologies are difficult to meet diverse application needs, specifically: (1) Inability to clearly define business requirements: Existing recognition technologies applied to operational businesses usually involve many types, requiring clear definition of business requirements. (2) High functional limitations: Existing technologies usually only support basic data monitoring and lack multi-functional applications such as intrusion detection alarms and remote real-time monitoring, limiting their use in complex scenarios. (3) Large size and inflexible deployment: Many existing devices, such as data acquisition servers, are large in size and not flexible in design, making them unsuitable for edge acquisition in narrow areas, thus limiting their application scope. (4) High cost: Traditional technologies usually result in high hardware and software investment costs, making it difficult to achieve efficient data integration and economical operation models.

[0030] In the operation of modern enterprises, a large number of complex business processes and data information are involved. How to conduct efficient and accurate operational control has become the key to enhancing the competitiveness of enterprises. Traditional operational control methods rely on manual experience and simple data statistical analysis, which have problems such as low efficiency, poor accuracy, and insufficient real-time performance, making it difficult to meet the growing demand for refined management from enterprises.

[0031] With the continuous development of science and technology, image recognition and deep learning technologies have made significant progress. Image recognition technology can process and analyze various image information to extract valuable features; deep learning technology has powerful data analysis and model training capabilities, enabling it to automatically learn patterns and rules from large amounts of data. However, research and practice on effectively integrating these two technologies for operational management are still relatively limited, failing to fully leverage their advantages to solve the challenges in operational management.

[0032] Based on this, in this example embodiment, an operation management method is first provided. This method can be implemented using a server or a terminal device. The terminals described in this disclosure can include mobile terminals such as mobile phones, tablets, laptops, PDAs, personal digital assistants (PDAs), portable media players (PMPs), navigation devices, wearable devices, smart bracelets, pedometers, and fixed terminals such as desktop computers. Figure 1 schematically illustrates a flowchart of an operation management method according to some embodiments of this disclosure. Referring to Figure 1, the operation control method may include the following steps: Step S110, acquiring operation image data, extracting features from the operation image data to obtain operation image feature vectors; Step S120, acquiring historical operation business data and a pre-built initial model, wherein the historical operation business data has temporal characteristics, and the initial model includes a long short-term memory network structure; Step S130, training the initial model based on the operation image feature vectors and historical operation business data to obtain an operation control prediction model; Step S140, acquiring real-time operation image data and real-time operation business data; Step S150, inputting the real-time operation image data and real-time operation business data into the operation control prediction model to obtain operation status data, which is used to determine the operation control scheme.

[0033] According to the operation control method in this example embodiment, on the one hand, key information in the operation process can be accurately extracted through image recognition, and combined with the powerful data analysis capabilities of the model, the enterprise's operating status can be predicted more accurately, reducing errors in human judgment and improving the accuracy of decision-making; on the other hand, by collecting and processing image and business data in real time, the model can quickly evaluate the operating situation, enabling relevant personnel to keep abreast of the enterprise's operational dynamics and respond in a timely manner, thus preventing the problem from escalating.

[0034] The operation control method in this example embodiment will be further explained below.

[0035] In one exemplary embodiment of this disclosure, step S110, acquiring operational image data and extracting features from the operational image data to obtain an operational image feature vector, includes: acquiring initial operational image data; performing image preprocessing on the initial operational image data to obtain operational image data; extracting features from the operational image data to obtain global and local features corresponding to the operational image data; and generating an operational image feature vector based on the global and local features.

[0036] The initial operational image data can be operational images and related data acquired through image acquisition equipment. Image preprocessing can be a preprocessing procedure involving denoising, grayscale conversion, and normalization of the acquired raw images. Operational image data can be image data obtained after image preprocessing the initial operational image data. Global features can be the overall attributes of the operational image data; common global features include, but are not limited to, color features, texture features, and shape features, such as intensity histograms. Local features can be features with local characteristics in the operational image data, such as, but not limited to, corner points and Speeded Up Robust Features (SURF). The operational image feature vector can be a vector representation of image features constructed based on the extracted global and local features.

[0037] After acquiring initial operational image data through image acquisition equipment, image preprocessing is performed on the raw image data. This includes denoising, grayscale conversion, and normalization to improve image quality and facilitate subsequent feature extraction and analysis. Following the acquisition of the operational image data, feature extraction is performed using image recognition techniques such as Convolutional Neural Networks (CNNs) to extract corresponding global and local features.

[0038] Furthermore, CNNs can automatically learn local and global features in images through structures such as convolutional layers, pooling layers, and fully connected layers, generating feature vector representations of the images, i.e., operational image feature vectors. By extracting features from images, key information in operational images can be accurately extracted, serving as the data foundation for subsequent model training.

[0039] In one exemplary embodiment of this disclosure, obtaining initial operational image data includes: determining the operational area to be collected based on operational control requirements, and configuring image acquisition devices in the operational area to be collected; and acquiring operational-related images of the operational area to be collected using the image acquisition devices as initial operational image data.

[0040] The operational area to be collected can be any area where image collection is required during the operational process. Operation-related images can be image data related to the company's operations.

[0041] This disclosure addresses the shortcomings of existing data acquisition servers, which are often bulky and lack flexibility. It describes the deployment of suitable image acquisition devices in different operational areas to collect image data generated during business operations. These devices, such as cameras and scanners, are distributed throughout various operational processes to gather relevant image information. Taking the daily operation of a sand quarry as an example, appropriate sensing devices and components can be selected for different scenarios, including ordinary river sections, cross-sections, centralized docking points at sand quarries / large-scale river and lake scenes, and sand pits themselves. This ensures that the site in each operational area is covered and image data is collected accordingly.

[0042] Referring to Figure 2, which schematically illustrates a scheme for deploying image acquisition equipment in special areas such as ordinary river sections or cross-sections according to an exemplary embodiment of this disclosure. For ordinary river sections, at least one monitoring point can be deployed every 1 kilometer, and black light / full color equipment is recommended. For wider river channels, points can be deployed across both banks, provided that power supply and grid connection are guaranteed. In addition, special attention should be paid to cross-sections / sand mining sections / banklines, and dual-spectrum equipment can be deployed at surrounding iron towers or other high points. For small river sections, a pole-mounted deployment method can also be used.

[0043] Referring to Figure 3, Figure 3 schematically illustrates a scheme for deploying image acquisition equipment at a centralized docking point for sand dredgers or in a large river or lake scene according to an exemplary embodiment of this disclosure. For sand dredging areas such as centralized docking points for sand dredgers or large river or lake scenes, panoramic monitoring equipment can be deployed at surrounding iron towers or other high points.

[0044] Referring to Figure 4, which schematically illustrates a scheme for deploying an image acquisition device in a sand field according to an exemplary embodiment of this disclosure. For the sand field area, dual-spectrum equipment can be deployed at surrounding towers or other high points, or at lower points on poles if the power supply is guaranteed. If there is a fixed access passage, checkpoint equipment can be deployed.

[0045] In addition, for other operational areas to be collected, including production lines, warehouses, and office areas, corresponding image acquisition devices can be used to collect operational data. This can include, but is not limited to, images of product appearance on the production line, images of goods arrangement in the warehouse, and images of documents in the office area. By using different sensing devices, accurate identification can be ensured even with obstructions. By collecting images from different operational areas, operational images of various stages of the enterprise can be obtained for subsequent model training.

[0046] For step S120, historical operational business data and a pre-built initial model are acquired. The historical operational business data has temporal characteristics, and the initial model includes a long short-term memory network structure. The historical operational business data can be relevant business data generated by the enterprise during its past operations. For example, historical operational business data may include, but is not limited to, business data generated in different operational stages within a specified past time period, as well as other related business data. Related business data may include, but is not limited to, sales data, financial data, and human resources data.

[0047] The system acquires operational data generated during past business operations as historical operational data. This historical operational data can be further collected and organized, with standardized data formats to prepare for subsequent data fusion. Additionally, a deep learning model is constructed; for example, in this embodiment, a deep learning model incorporating Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTMs) can be built as the initial model.

[0048] In one exemplary embodiment of this disclosure, step S130 involves training an initial model using a Long Short-Term Memory (LSTM) network structure based on the time-series features of the model input data to obtain an operation management prediction model. This includes: sorting historical operation business data according to chronological order to obtain operation business sequence data; labeling the operation business sequence data to obtain labeled operation business data; fusing the labeled operation business data with operation image feature vectors to obtain fused input data; determining the operation prediction result corresponding to the fused input data using the initial model; and training the initial model based on a pre-constructed loss function, the operation prediction result, and the historical operation business data until the training termination condition is met to obtain the operation management prediction model.

[0049] The model input data can be operational image feature vectors and historical operational business data. Operational business sequence data can be obtained by serializing historical operational business data in chronological order. Annotated operational business data can be annotated operational business sequence data; for example, operational business sequence data and corresponding operational image feature vectors can be uniformly annotated and used as model training data. Fusion input data can be the input data obtained by fusing annotated operational business data and operational image feature vectors. The operational management and control prediction model can be a network model that predicts and analyzes relevant indicator data of the enterprise's operational status.

[0050] During model training, operational image feature vectors and historical operational business data are used as the initial model input data. Before inputting these into the initial model, the historical operational business data can be sorted chronologically to obtain corresponding operational business sequence data. This operational business sequence data is then labeled to obtain labeled operational business data. When the labeled operational business data and operational image feature vectors are input into the initial model, the model's data fusion module can fuse them as fused input data.

[0051] The fused input data undergoes feature extraction and learning using the network structures in the initial model. For example, the RNN and LSTM structures in the initial model can handle sequential data, learning the time-series features and long-term dependencies in the fused input data. For a large amount of model input data, this data is fed into the initial model, which then outputs the corresponding operational prediction result based on the learned time-series features and long-term dependencies. After obtaining the operational prediction result, a loss function is used to calculate the difference between the operational prediction result and historical operational business data.

[0052] The core logic of the initial model in processing sequence data is as follows: For labeled operational business data, since this data is sequence data, let's assume the labeled operational business data is [x1, x2, ..., x...]. t ], where t is the time step, and the RNN will compute the hidden state at each time t. The formula for the hidden state is: = tanh( + + ),in, It is the hidden state of the previous moment, which means that the current information will "inherit" the result of the previous moment, thus connecting the entire sequence; It is the current moment. The calculation weights; yes The calculation weights; It is a bias term.

[0053] The specific method for learning time series features is: hidden state Real-time integration of current input and historical information Therefore, it can capture short-term associations in a sequence. For example, when processing text, RNNs can determine the semantics of the current word based on the information of the previous few words; when processing time-series data (such as operational business data), it can capture the changing trends of business data at adjacent time points.

[0054] When a sequence is too long (e.g., more than 20 time steps), calculating the gradient involves multiple matrix multiplications. Since the absolute value of the derivative of an activation function (e.g., tanh) is less than 1, the gradient will gradually approach 0 after multiple multiplications, causing the model to be unable to update the parameters of earlier time steps and thus "forget" the key information from earlier parts of the long sequence. At this point, the LSTM structure adds cell states and three gating units to the RNN to specifically address the gradient vanishing problem, enabling the preservation and control of long-term information.

[0055] Specifically, the core structure includes: cell states and three gates. Cell states act like an "information conveyor belt," directly passing information between time steps. They are fine-tuned only through gating, avoiding frequent gradient decay and are crucial for preserving long-term information. The three gates—forget gate, input gate, and output gate—control the "passing through / blocking" of information using a sigmoid activation function (outputting a value between 0 and 1), thus filtering information. In long sequences, useful early information (such as the "initial trend" in time-series data) is directly passed through the cell states. The forget gate only filters noise, while the input gate continuously replenishes new key information. Because the gradient decay of the cell states is extremely slow, the model can effectively update parameters from earlier time steps, thereby "remembering" information from dozens or even hundreds of time steps ago, overcoming the short-term dependency limitation of RNNs.

[0056] The initial model is continuously trained based on a loss function until the loss value between the operational forecast and historical operational data is minimized, or until the training iterations reach a specified number, at which point training stops, resulting in the operational management and prediction model. For example, the model parameters can be continuously adjusted using a backpropagation algorithm to ensure that the model's operational forecasts closely approximate the actual operational situation. During training, methods such as cross-validation can also be employed to ensure the model's generalization ability and stability.

[0057] For step S150, real-time operational image data and real-time operational business data are input into the operational control prediction model to obtain operational status data, which is used to determine the operational control plan.

[0058] Real-time operational image data can be images of various operational areas acquired in real time through image acquisition equipment. Real-time operational business data can be business data related to the company's operational status acquired in real time.

[0059] In order to monitor the operational status of an enterprise, this embodiment inputs real-time operational image data and real-time operational business data into a trained operational control prediction model, so that the operational control prediction model can predict the operational status of the enterprise in real time and output operational status data.

[0060] In one exemplary embodiment of this disclosure, the model prediction results corresponding to the operation management prediction model and the operation feedback results in the business operation scenario are obtained; the model optimization target is determined based on the model prediction results and the operation feedback results; the operation management prediction model is optimized and adjusted based on the model optimization target to obtain the adjusted operation management prediction model. The optimization and adjustment operations include one or more of the following: adjusting the model structure, increasing the amount of training data, and optimizing the model parameters.

[0061] The model prediction results can be operational status data obtained by the operation management prediction model based on data analysis and prediction of real-time operational image data and real-time operational business data. Business operation scenarios can be various scenarios in the enterprise's operation process. Operational feedback results can be feedback generated by relevant personnel after comparing the model prediction results with the actual operational status data of the enterprise and evaluating the accuracy of the model prediction results. Model optimization objectives can be the goals that the model aims to achieve after optimization and adjustment of the operation management prediction model. For example, model optimization objectives can include, but are not limited to, the specific accuracy level the model aims to achieve, or the performance improvement goals of the model.

[0062] After obtaining the model's prediction results and operational feedback, and determining the model optimization objectives based on these results, the operational management prediction model is optimized and adjusted accordingly. This optimization may involve adjusting the model structure, increasing the amount of training data, or optimizing model parameters—using one or more of these methods—to improve the model's performance and obtain an adjusted operational management prediction model. This aims to improve the model's prediction accuracy and efficiency. Simultaneously, regularization methods are introduced to prevent overfitting.

[0063] Through the above model optimization and adjustment steps, the adjusted model can adapt to complex and ever-changing operating environments: deep learning models have good learning and adaptability, and can learn complex operating patterns and rules from a large amount of historical data. Even when the operating environment changes, they can continuously provide effective operational management support for enterprises through continuous learning and optimization.

[0064] In one exemplary embodiment of this disclosure, a pre-configured operational status threshold is obtained; the operational status data is compared with the operational status threshold to obtain the operational indicator difference; when the operational indicator difference is not within the difference deviation range, operational early warning information is generated; or a pre-built operational early warning analysis model is obtained; the operational early warning analysis model analyzes the operational status data to generate operational early warning information.

[0065] The operational status threshold can be a reference value compared with the operational status data. The operational indicator difference can be the difference between the operational status data and the operational status threshold. The difference deviation range can be the data range corresponding to the acceptable operational indicator difference. The operational early warning information can be a systematic information system that triggers alarms in real time after analyzing the operational status data and identifying anomalies in key indicators. The operational early warning analysis model can be a model that generates operational early warning information based on the obtained trend information after analyzing and processing the operational status data.

[0066] The operational management prediction model, after predicting and analyzing the enterprise's operational status based on real-time data input from the model, can obtain corresponding operational status data. The operational status data output by the model is compared with pre-configured operational status thresholds, and the difference between the two is calculated to obtain the operational indicator difference.

[0067] For the identified operational indicator discrepancies, it is determined whether these discrepancies fall within the deviation range. If they do not, the company's operational status indicators are considered to have failed to meet expected standards. In this case, corresponding operational warning information can be generated based on the discrepancies that are outside the deviation range. Alternatively, operational status warnings can be generated using a pre-built operational warning analysis model. For example, operational status data output from an operational control prediction model can be input into the operational warning analysis model, which will then analyze the data and generate corresponding operational warning information. Through these data analysis steps, when operational indicators are found to deviate from the normal range, timely warning information is issued to remind managers to take appropriate measures to prevent the problem from escalating.

[0068] In one exemplary embodiment of this disclosure, operational status data is compared and analyzed with historical operational business data to obtain operational status change trend information; based on the operational status change trend information, an operational control plan is generated.

[0069] Among these, operational status change trend information can be relevant information reflecting changes in the company's operational status and potential future trends in operational data. This information can be determined by comparing and analyzing the company's current operational status data with historical operational status data. Operational control plans can be specific decision-making schemes for managing and controlling the company's operations.

[0070] By comparing and analyzing the real-time operational status data output by the operational control prediction model with historical operational business data, the trend of operational status changes from a certain point in the past to the present can be determined. Based on these changes, a matching operational control plan can be formulated. Specifically, based on the model's prediction results and the analysis of historical data, decision support is provided to enterprise managers. For example, in production planning, reasonable production scheduling suggestions are given based on product demand forecasts and production capacity analysis; in inventory management, suggestions for optimizing inventory strategies are proposed based on sales trend forecasts and inventory levels. This disclosure reduces the workload of manual statistics and analysis through automated data processing and analysis, improving the efficiency of operational control. Simultaneously, model-based decision suggestions can optimize business processes, further enhancing the overall operational efficiency of the enterprise.

[0071] In summary, the operational control method disclosed herein involves: acquiring operational image data; extracting features from the operational image data to obtain operational image feature vectors; acquiring historical operational business data and a pre-built initial model, where the historical operational business data has temporal characteristics and the initial model includes a long short-term memory network structure; training the initial model based on the operational image feature vectors and historical operational business data to obtain an operational control prediction model; acquiring real-time operational image data and real-time operational business data; and inputting the real-time operational image data and real-time operational business data into the operational control prediction model to obtain operational status data, which is used to determine operational control plans. On the one hand, by accurately extracting key information in the operational process through image recognition and combining it with the powerful data analysis capabilities of the model, the operational status of the enterprise can be predicted more accurately, reducing errors in human judgment and improving the accuracy of decision-making. On the other hand, by collecting and processing image and business data in real time, the model can quickly assess the operational situation, enabling relevant personnel to promptly grasp the dynamics of enterprise operations, respond in a timely manner, and prevent the escalation of problems. On the other hand, deep learning models possess excellent learning and adaptability, enabling them to learn complex operational patterns and rules from vast amounts of historical data. Even when the operational environment changes, they can continuously learn and optimize, providing effective operational management support for enterprises. Furthermore, automated data processing and analysis reduce the workload of manual statistics and analysis, improving the efficiency of operational management. Simultaneously, model-based decision recommendations can optimize business processes, further enhancing the overall operational efficiency of the enterprise.

[0072] It should be noted that although the steps of the method in this invention are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0073] Furthermore, in this example embodiment, an operation management and control device is also provided. Referring to FIG5, the operation management and control device 500 may include: an image feature extraction module 510, a data model acquisition module 520, a model training module 530, a real-time data acquisition module 540, and an operation status prediction module 550.

[0074] Specifically, the image feature extraction module 510 is used to acquire operational image data, extract features from the operational image data, and obtain operational image feature vectors; the data model acquisition module 520 is used to acquire historical operational business data and a pre-built initial model. The historical operational business data has temporal characteristics, and the initial model includes a long short-term memory network structure; the model training module 530 is used to train the initial model based on the operational image feature vectors and historical operational business data to obtain an operational control prediction model; the real-time data acquisition module 540 is used to acquire real-time operational image data and real-time operational business data; and the operational status prediction module 550 is used to input the real-time operational image data and real-time operational business data into the operational control prediction model to obtain operational status data, which is used to determine the operational control plan.

[0075] In one exemplary embodiment of this disclosure, the image feature extraction module 510 includes an image feature extraction unit, configured to: acquire initial operational image data; perform image preprocessing on the initial operational image data to obtain operational image data; extract features from the operational image data to obtain global features and local features corresponding to the operational image data; and generate an operational image feature vector based on the global features and local features.

[0076] In one exemplary embodiment of this disclosure, the image feature extraction unit includes an image acquisition subunit, configured to: determine the operational area to be collected based on operational control requirements, and configure an image acquisition device in the operational area to be collected; and acquire operational-related images of the operational area to be collected using the image acquisition device as initial operational image data.

[0077] In one exemplary embodiment of this disclosure, the model input data includes operational image feature vectors and historical operational business data. The model training module 530 includes a model training unit, used for: labeling the historical operational business data to obtain labeled operational business data; sorting the labeled operational business data according to chronological order to obtain labeled operational business data; fusing the labeled operational business data and operational image feature vectors to obtain fused input data; determining the operational prediction result corresponding to the fused input data using an initial model; and training the initial model based on a pre-constructed loss function, the operational prediction result, and the historical operational business data until the training termination condition is met to obtain an operational control prediction model.

[0078] In one exemplary embodiment of this disclosure, the operation management and control device 500 further includes a model optimization module, configured to: obtain the model prediction result corresponding to the operation management and control prediction model, and the operation feedback result in the business operation scenario; determine the model optimization target based on the model prediction result and the operation feedback result; and perform optimization and adjustment operations on the operation management and control prediction model based on the model optimization target to obtain the adjusted operation management and control prediction model, wherein the optimization and adjustment operations include one or more of adjusting the model structure, increasing the amount of training data, and optimizing the model parameters.

[0079] In one exemplary embodiment of this disclosure, the operation control device 500 further includes an early warning information generation module, used for: acquiring a pre-configured operation status threshold; comparing the operation status data with the operation status threshold to obtain an operation indicator difference; generating operation early warning information when the operation indicator difference is not within the difference deviation range; or acquiring a pre-built operation early warning analysis model; and analyzing the operation status data using the operation early warning analysis model to generate operation early warning information.

[0080] In one exemplary embodiment of this disclosure, the operation control device 500 further includes a control scheme generation module, used to: compare and analyze the operation status data with historical operation business data to obtain operation status change trend information; and generate an operation control scheme based on the operation status change trend information.

[0081] The specific details of the virtual modules of each of the above-mentioned operation and control devices have been described in detail in the corresponding operation and control methods. For any undisclosed details, please refer to the implementation methods in the method section, and therefore will not be repeated here.

[0082] It should be noted that although several modules or units of the operation and control device have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0083] An exemplary embodiment of this disclosure also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the above-described operation control method.

[0084] In one embodiment, the computer program product may be a tangible product containing a computer program, such as a computer-readable storage medium storing the computer program. Referring to FIG6, FIG6 schematically illustrates a computer-readable storage medium 600 according to an exemplary embodiment of the present disclosure. The computer-readable storage medium 600 may be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory, hard disk drive (HDD), solid-state drive (SSD), etc. Exemplarily, the computer program product may be implemented as a non-volatile storage medium storing a computer program, such as read-only memory, NAND flash memory, etc.

[0085] In one implementation, the computer program product can be an intangible product containing a computer program. For example, the computer program product can be implemented as a virtual digital product, such as an executable file, installation package, or other digital file storing the computer program.

[0086] Computer program code can be written in one or more programming languages. Examples of programming languages ​​include C, Java, and C++. Program code can execute entirely on the user's computing device, partially on the user's computing device, or as a standalone software package. It can also execute partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, such as a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via an internet connection provided by a mobile network operator).

[0087] Computer programs can be carried or transmitted via signals such as electricity, magnetism, light, electromagnetic fields, and infrared radiation. Electronic devices can convert signals carrying computer programs into digital signals, thereby running the computer programs. When a computer program runs on an electronic device, its code is used to cause the electronic device to execute (more specifically, to execute by the processor of the electronic device) the method steps of various exemplary embodiments of this disclosure, such as the operation control method described above.

[0088] Exemplary embodiments of this disclosure also provide an electronic device, which may include a processor and a memory. The memory stores executable instructions of the processor, such as a computer program. The processor executes the executable instructions to perform the method steps of various exemplary embodiments of this disclosure. Furthermore, the electronic device may also include a display for displaying a graphical user interface.

[0089] Referring now to FIG7, an electronic device will be described by way of example in the form of a general computing device. It should be understood that the electronic device 700 shown in FIG7 is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of this disclosure.

[0090] As shown in Figure 7, the electronic device 700 may include: a processor 710, a memory 720, a bus 730, an I / O (input / output) interface 740, a network adapter 750, and a display 760.

[0091] The memory 720 may include volatile memory, such as RAM 721 and cache unit 722, and may also include non-volatile memory, such as ROM 723. The memory 720 may also include one or more program modules 724, including but not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. For example, program module 724 may include the modules described above.

[0092] The processor 710 may include one or more processing units, such as an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor, and / or an NPU (Neural-Network Processing Unit).

[0093] The processor 710 can be used to execute executable instructions stored in the memory 720, such as the aforementioned operation control method.

[0094] Bus 730 is used to connect different components of electronic device 700 and may include a data bus, an address bus and a control bus.

[0095] Electronic device 700 can communicate with one or more external devices 800 (such as keyboard, mouse, external controller, etc.) through I / O interface 740.

[0096] Electronic device 700 can communicate with one or more networks via network adapter 750. For example, network adapter 750 can provide mobile communication solutions such as 3G / 4G / 5G, or wireless communication solutions such as wireless LAN, Bluetooth, and near-field communication. Network adapter 750 can communicate with other modules of electronic device 700 via bus 730.

[0097] The electronic device 700 can display a graphical user interface through the display 760, such as displaying operational status data and an interface for showcasing operational control plans.

[0098] Although not shown in Figure 7, other hardware and / or software modules may be configured in electronic device 700, including but not limited to: display (if electronic device includes display as mentioned above, then "display" is removed here), microcode, device driver, redundant processor, external disk drive array, RAID system, tape drive, and data backup storage system, etc.

[0099] It should be understood that this disclosure is not limited to the specific methods, steps, or structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. Those skilled in the art will readily conceive of other embodiments based on the specific implementations provided in this disclosure. Therefore, the specific implementations provided in this disclosure are merely exemplary, and the scope and spirit of this disclosure are indicated by the claims, and should cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary technical means in the art not disclosed in this disclosure.

Claims

1. An operational control method, characterized in that, include: The process involves: acquiring operational image data; extracting features from the operational image data to obtain operational image feature vectors; acquiring historical operational business data and a pre-built initial model, wherein the historical operational business data has temporal characteristics and the initial model includes a long short-term memory network structure; training the initial model based on the operational image feature vectors and the historical operational business data to obtain an operational management and control prediction model; acquiring real-time operational image data and real-time operational business data; and inputting the real-time operational image data and real-time operational business data into the operational management and control prediction model to obtain operational status data, which is used to determine operational management and control schemes.

2. The method according to claim 1, characterized in that, The step of acquiring operational image data and extracting features from the operational image data to obtain an operational image feature vector includes: acquiring initial operational image data; performing image preprocessing on the initial operational image data to obtain the operational image data; extracting features from the operational image data to obtain global and local features corresponding to the operational image data; and generating the operational image feature vector based on the global and local features.

3. The method according to claim 2, characterized in that, The process of acquiring initial operational image data includes: determining the operational area to be collected based on operational management and control requirements, and configuring image acquisition devices in the operational area to be collected; acquiring operational-related images of the operational area to be collected through the image acquisition devices, which are used as the initial operational image data.

4. The method according to claim 1, characterized in that, The initial model further includes a recurrent neural network. Training the initial model based on the operational image feature vector and the historical operational business data to obtain an operational management and control prediction model includes: sorting the historical operational business data according to chronological order to obtain operational business sequence data; labeling the operational business sequence data to obtain labeled operational business data; fusing the labeled operational business data with the operational image feature vector to obtain fused input data; determining the operational prediction result corresponding to the fused input data using the initial model; and training the initial model based on a pre-constructed loss function, the operational prediction result, and the historical operational business data until the training termination condition is met to obtain the operational management and control prediction model.

5. The method according to claim 4, characterized in that, The method further includes: obtaining the model prediction results corresponding to the operation management and control prediction model, and the operation feedback results in the business operation scenario; determining the model optimization target based on the model prediction results and the operation feedback results; and performing optimization and adjustment operations on the operation management and control prediction model based on the model optimization target to obtain the adjusted operation management and control prediction model, wherein the optimization and adjustment operations include one or more of adjusting the model structure, increasing the amount of training data, and optimizing the model parameters.

6. The method according to claim 1, characterized in that, The method further includes: obtaining a pre-configured operational status threshold; comparing the operational status data with the operational status threshold to obtain an operational indicator difference; generating operational early warning information when the operational indicator difference is not within the difference deviation range; or obtaining a pre-built operational early warning analysis model; and analyzing the operational status data using the operational early warning analysis model to generate operational early warning information.

7. The method according to claim 1, characterized in that, The method further includes: comparing and analyzing the operational status data with historical operational business data to obtain operational status change trend information; and generating the operational management and control plan based on the operational status change trend information.

8. An operation control device, characterized in that, include: The image feature extraction module is used to acquire operational image data and extract features from the operational image data to obtain operational image feature vectors; the data model acquisition module is used to acquire historical operational business data and a pre-built initial model, wherein the historical operational business data has temporal characteristics and the initial model includes a long short-term memory network structure. The model training module is used to train the initial model based on the operational image feature vector and the historical operational business data to obtain an operational management and control prediction model. The real-time data acquisition module is used to acquire real-time operational image data and real-time operational business data; the operational status prediction module is used to input the real-time operational image data and the real-time operational business data into the operational control prediction model to obtain operational status data, which is used to determine the operational control plan.

9. An electronic device, characterized in that, include: processor; A memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the operation control method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the operation control method according to any one of claims 1 to 7.

Images