Multi-mode AIGC generation service control method and system

Abstract

The invention provides a multi-modal AIGC generation service control method and system, and the method comprises the steps: selecting processing components for different types of contents from an AIGC library, enabling all selected processing components to form a multi-modal component set, loading the multi-modal component set to a user side based on the real-time working state information of the user side, and carrying out the real-time operation of the user side. Therefore, the user side can carry out multi-mode AIGC service; based on execution state information of an AI content generation task of a user side, a corresponding processing component is called from a multi-modal component set to complete a corresponding AI content generation thread, so that different modal processing components are flexibly and quickly used in an AI content generation process, and the generation reliability and efficiency of the AI content are improved; and when the AI content generation task is correctly executed, outputting the final generation content of the AI content generation task, and performing integration and storage operation on the multi-modal component set currently loaded by the user side, so that the complexity of AI content generation is reduced, and one-stop AIGC content generation is realized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a multimodal AIGC generation service control method and system. Background Art

[0002] Artificial Intelligence Generated Content (AIGC) is usually a single AIGC service based on a deep learning algorithm platform. It is unable to integrate AIGC services of different modalities such as text, images, videos and voices in a collaborative mode. In the process of generating corresponding AI content, it is necessary to load AIGC services of a single modality separately on the deep learning algorithm platform, which increases the complexity of AI content generation and cannot achieve one-stop AIGC content integration generation. Summary of the invention

[0003] The purpose of the present invention is to provide a multimodal AIGC generation service control method and system, which selects processing components for different types of content from an AIGC library, and forms a multimodal component set with all the selected processing components, and loads the multimodal component set to the user end based on the real-time working status information of the user end, so that the user end can perform multimodal AIGC services; based on the execution status information of the AI ​​content generation task of the user end, the corresponding processing component is called from the multimodal component set to complete the corresponding AI content generation thread, so that different modal processing components can be used flexibly and quickly in the AI ​​content generation process, thereby improving the generation reliability and efficiency of AI content; and when the AI ​​content generation task is correctly executed, the final generated content of the AI ​​content generation task is output, and the multimodal component set currently loaded by the user end is integrated and saved, so that the user end can reuse the multimodal component set for AIGC services in the future, reduce the complexity of AI content generation, and realize one-stop AIGC content generation.

[0004] The present invention is achieved through the following technical solutions:

[0005] A multi-modal AIGC generation service control method, comprising:

[0006] Based on a content generation request from a user terminal, processing components for different types of content are selected from the AIGC library, and all the selected processing components are combined into a multimodal component set; based on real-time working status information of the user terminal, the multimodal component set is loaded into the user terminal, and a multimodal component thread matching operation is performed on the user terminal;

[0007] Obtaining execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, calling a corresponding processing component from the multimodal component set to complete a corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, updating a corresponding processing component in the multimodal component set, thereby adjusting the execution status of the AI ​​content generation thread;

[0008] The final generated content of the AI ​​content generation task is detected to determine whether the AI ​​content generation task is executed correctly; when the AI ​​content generation task is executed correctly, the final generated content is output, and the multimodal component set currently loaded on the user end is integrated and saved.

[0009] Optionally, based on a content generation request from a user terminal, processing components for different types of content are selected from the AIGC library, and all selected processing components are combined into a multimodal component set; based on real-time working status information of the user terminal, the multimodal component set is loaded into the user terminal, and a multimodal component thread matching operation is performed on the user terminal, including:

[0010] Parse and process the content generation request from the user end to determine the modal attribute information of the AI ​​content that the user end currently expects to generate; based on the modal attribute information, select processing components for at least two of the text modality, image modality, video modality, and voice modality from the AIGC library, and form all the selected processing components into a multimodal component set;

[0011] Based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined; based on the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power.

[0012] Optionally, obtaining the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, calling the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, updating the corresponding processing component in the multimodal component set, thereby adjusting the execution status of the AI ​​content generation thread, includes:

[0013] Obtaining execution progress information of an AI content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, thereby determining the AI ​​content generation subtask currently executed by the user terminal; based on the currently executed AI content generation subtask, calling a corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask;

[0014] The execution speed of the AI ​​content generation thread is compared with a preset execution speed threshold. If the execution speed is less than the preset execution speed threshold, the corresponding processing component in the multimodal component set is updated to increase the execution speed of the AI ​​content generation thread.

[0015] Optionally, according to the relationship between the execution speed and a preset execution speed threshold, performing abnormality determination on the operation of the AI ​​content generation thread includes:

[0016] If the execution speed is not less than the preset execution speed threshold, monitor the first operating parameter of the AI ​​content generation thread in real time, wherein the first operating parameter includes the amount of content data generated per unit time, memory utilization, the data amount difference between the input content and the output content, and the content generation bit error rate;

[0017] Obtaining a first operation evaluation coefficient using the content data volume generated per unit time, memory utilization, data volume difference between input content and output content, and content generation bit error rate included in the first operation parameter;

[0018] The first operation evaluation coefficient is obtained by the following formula:

[0019]

[0020] Among them, R 01 represents the first running evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; M i represents the memory utilization corresponding to the i-th unit time; E i represents the content generation bit error rate corresponding to the i-th unit time;

[0021] comparing the first operation evaluation coefficient with a preset coefficient threshold;

[0022] When the first operation evaluation coefficient exceeds a preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued;

[0023] If the execution speed is less than the preset execution speed threshold, the second operation parameter of the AI ​​content generation thread is monitored in real time, wherein the second operation parameter includes the amount of content data generated per unit time, the difference in data volume between the input content and the output content, and the execution speed value;

[0024] Obtaining a second operation evaluation coefficient using the content data volume generated per unit time, the data volume difference between the input content and the output content, and the execution speed value included in the second operation parameter;

[0025] The second operation evaluation coefficient is obtained by the following formula:

[0026]

[0027] Among them, R 02 represents the second operation evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; C zmaxi represents the maximum value of the data volume difference between the input content and the output content corresponding to the i-th unit time; v i represents the execution speed value generated corresponding to the i-th unit time; v b Indicates the guaranteed error of the generated execution speed value corresponding to n unit time;

[0028] comparing the second operation evaluation coefficient with a preset coefficient threshold;

[0029] When the second operation evaluation coefficient exceeds the preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued.

[0030] Optionally, the final generated content of the AI ​​content generation task is detected to determine whether the AI ​​content generation task is correctly executed; when the AI ​​content generation task is correctly executed, the final generated content is output, and the multimodal component set currently loaded by the user terminal is integrated and saved, including:

[0031] Detecting the final generated content of the AI ​​content generation task to obtain a data bit error rate of the final generated content; if the data bit error rate is greater than or equal to a preset bit error rate threshold, determining that the AI ​​content generation task is not correctly executed; otherwise, determining that the AI ​​content generation task is correctly executed;

[0032] When the AI ​​content generation task is executed correctly, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user terminal is integrated into a repeatedly callable multimodal component set and saved in a designated storage space inside the user terminal.

[0033] Multimodal AIGC generation service control system, including:

[0034] A multimodal component determination module, configured to select processing components for different types of content from the AIGC library based on a content generation request from a user terminal, and to form a multimodal component set from all selected processing components;

[0035] A multimodal component matching module, used to load the multimodal component set into the user terminal based on the real-time working status information of the user terminal, and perform a multimodal component thread matching operation on the user terminal;

[0036] The AI ​​content generation task processing module is used to obtain the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, update the corresponding processing component in the multimodal component set to adjust the execution status of the AI ​​content generation thread;

[0037] An AI content generation task judgment module, used to detect the final generated content of the AI ​​content generation task and determine whether the AI ​​content generation task is executed correctly;

[0038] The multimodal component set operation module is used to output the final generated content when the AI ​​content generation task is executed correctly, and to integrate and save the multimodal component set currently loaded on the user end.

[0039] Optionally, the multimodal component determination module is used to select processing components for different types of content from the AIGC library based on a content generation request from the user end, and form a multimodal component set with all the selected processing components, including:

[0040] Parse and process the content generation request from the user end to determine the modal attribute information of the AI ​​content that the user end currently expects to generate; based on the modal attribute information, select processing components for at least two of the text modality, image modality, video modality, and voice modality from the AIGC library, and form all the selected processing components into a multimodal component set;

[0041] The multimodal component matching module is used to load the multimodal component set to the user terminal based on the real-time working status information of the user terminal, and perform a multimodal component thread matching operation on the user terminal, including:

[0042] Based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined; based on the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power.

[0043] Optionally, the AI ​​content generation task processing module is used to obtain the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, update the corresponding processing component in the multimodal component set to adjust the execution status of the AI ​​content generation thread, including:

[0044] Obtaining execution progress information of an AI content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, thereby determining the AI ​​content generation subtask currently executed by the user terminal; based on the currently executed AI content generation subtask, calling a corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask;

[0045] The execution speed of the AI ​​content generation thread is compared with a preset execution speed threshold. If the execution speed is less than the preset execution speed threshold, the corresponding processing component in the multimodal component set is updated to increase the execution speed of the AI ​​content generation thread.

[0046] Optionally, according to the relationship between the execution speed and a preset execution speed threshold, performing abnormality determination on the operation of the AI ​​content generation thread includes:

[0047] If the execution speed is not less than the preset execution speed threshold, monitor the first operating parameter of the AI ​​content generation thread in real time, wherein the first operating parameter includes the amount of content data generated per unit time, memory utilization, the data amount difference between the input content and the output content, and the content generation bit error rate;

[0048] Obtaining a first operation evaluation coefficient using the content data volume generated per unit time, memory utilization, data volume difference between input content and output content, and content generation bit error rate included in the first operation parameter;

[0049] The first operation evaluation coefficient is obtained by the following formula:

[0050]

[0051] Among them, R 01 represents the first running evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; M i represents the memory utilization corresponding to the i-th unit time; E i represents the content generation bit error rate corresponding to the i-th unit time;

[0052] comparing the first operation evaluation coefficient with a preset coefficient threshold;

[0053] When the first operation evaluation coefficient exceeds a preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued;

[0054] If the execution speed is less than the preset execution speed threshold, the second operation parameter of the AI ​​content generation thread is monitored in real time, wherein the second operation parameter includes the amount of content data generated per unit time, the difference in data volume between the input content and the output content, and the execution speed value;

[0055] Obtaining a second operation evaluation coefficient using the content data volume generated per unit time, the data volume difference between the input content and the output content, and the execution speed value included in the second operation parameter;

[0056] The second operation evaluation coefficient is obtained by the following formula:

[0057]

[0058] Among them, R 02 represents the second operation evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; C zmaxi represents the maximum value of the data volume difference between the input content and the output content corresponding to the i-th unit time; v i represents the execution speed value generated corresponding to the i-th unit time; v b Indicates the guaranteed error of the generated execution speed value corresponding to n unit time;

[0059] comparing the second operation evaluation coefficient with a preset coefficient threshold;

[0060] When the second operation evaluation coefficient exceeds the preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued.

[0061] Optionally, the AI ​​content generation task judgment module is used to detect the final generated content of the AI ​​content generation task to determine whether the AI ​​content generation task is correctly executed, including:

[0062] Detecting the final generated content of the AI ​​content generation task to obtain a data bit error rate of the final generated content; if the data bit error rate is greater than or equal to a preset bit error rate threshold, determining that the AI ​​content generation task is not correctly executed; otherwise, determining that the AI ​​content generation task is correctly executed;

[0063] The multimodal component set operation module is used to output the final generated content when the AI ​​content generation task is correctly executed, and to integrate and save the multimodal component set currently loaded by the user end, including:

[0064] When the AI ​​content generation task is executed correctly, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user terminal is integrated into a repeatedly callable multimodal component set and saved in a designated storage space inside the user terminal.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] The multimodal AIGC generation service control method and system provided in the present application select processing components for different types of content from the AIGC library, and form all the selected processing components into a multimodal component set, and based on the real-time working status information of the user end, load the multimodal component set to the user end, so that the user end can perform multimodal AIGC services; based on the execution status information of the AI ​​content generation task of the user end, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread, so that different modal processing components can be used flexibly and quickly in the AI ​​content generation process, thereby improving the reliability and efficiency of AI content generation; and when the AI ​​content generation task is correctly executed, the final generated content of the AI ​​content generation task is output, and the multimodal component set currently loaded on the user end is integrated and saved, so that the user end can reuse the multimodal component set for AIGC services in the future, reduce the complexity of AI content generation, and realize one-stop AIGC content generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0068] Figure 1 A flowchart of the multi-modal AIGC generation service control method provided by the present invention.

[0069] Figure 2 A schematic diagram of the structure of the multi-modal AIGC generation service control system provided by the present invention. DETAILED DESCRIPTION

[0070] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. It is to be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some structures related to the present application are shown in the accompanying drawings, rather than all structures. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0071] The terms "including" and "having" and any variations thereof in this application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0072] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0073] See also Figure 1 As shown, a multi-modal AIGC generation service control method provided in an embodiment of the present application includes:

[0074] Based on the content generation request from the user terminal, processing components for different types of content are selected from the AIGC library, and all the selected processing components are combined into a multimodal component set; based on the real-time working status information of the user terminal, the multimodal component set is loaded into the user terminal, and a multimodal component thread matching operation is performed on the user terminal;

[0075] Obtaining the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, calling the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, updating the corresponding processing component in the multimodal component set, thereby adjusting the execution status of the AI ​​content generation thread;

[0076] The final generated content of the AI ​​content generation task is detected to determine whether the AI ​​content generation task is executed correctly; when the AI ​​content generation task is executed correctly, the final generated content is output, and the multimodal component set currently loaded on the user end is integrated and saved.

[0077] The beneficial effects of the above embodiments are as follows: the multimodal AIGC generation service control method selects processing components for different types of content from the AIGC library, and forms a multimodal component set with all the selected processing components, and loads the multimodal component set to the user terminal based on the real-time working status information of the user terminal, so that the user terminal can perform multimodal AIGC services; based on the execution status information of the AI ​​content generation task of the user terminal, the corresponding processing component is called from the multimodal component set to complete the corresponding AI content generation thread, so that different modal processing components can be used flexibly and quickly in the AI ​​content generation process, thereby improving the reliability and efficiency of AI content generation; and when the AI ​​content generation task is correctly executed, the final generated content of the AI ​​content generation task is output, and the multimodal component set currently loaded by the user terminal is integrated and saved, so that the user terminal can reuse the multimodal component set for AIGC services in the future, reduce the complexity of AI content generation, and realize one-stop AIGC content generation.

[0078] In another embodiment, based on a content generation request from a user terminal, processing components for different types of content are selected from the AIGC library, and all the selected processing components are combined into a multimodal component set; based on the real-time working status information of the user terminal, the multimodal component set is loaded into the user terminal, and a multimodal component thread matching operation is performed on the user terminal, including:

[0079] Parse and process the content generation request from the user end to determine the modal attribute information of the AI ​​content that the user end currently expects to generate; based on the modal attribute information, select processing components for at least two of the text modality, image modality, video modality, and voice modality from the AIGC library, and form all the selected processing components into a multimodal component set;

[0080] Based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined; based on the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power.

[0081] The beneficial effect of the above embodiment is that when the user terminal needs to obtain AIGC services and generate corresponding AI content, the content generation request from the user terminal is first parsed and processed to determine the modal attribute information of the AI ​​content that the user terminal currently expects to generate. The modal attribute information may include but is not limited to text modal attribute information, image modal attribute information, video modal attribute information, and voice modal attribute information, so that the AI ​​content that the user terminal currently expects to generate can be comprehensively identified in terms of modality. Then, based on the modal attribute information, processing components for at least two of the text modality, image modality, video modality, and voice modality are selected from the AIGC library, and all the selected processing components are combined into a multimodal component set, so that the formed multimodal component set can provide matching multimodal AIGC services for the user terminal. In order to ensure the normal formation of the multimodal component set inside the user terminal, it is necessary to run the multimodal component set with the help of the threads in the user terminal. At this time, based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined, and then according to the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power, so that the multimodal component set can be fully triggered and run inside the user terminal.

[0082] In another embodiment, the execution state information of the AI ​​content generation task of the user terminal is obtained, and based on the execution state information, a corresponding processing component is called from the multimodal component set to complete the corresponding AI content generation thread; based on the execution state information of the AI ​​content generation thread, the corresponding processing component in the multimodal component set is updated to adjust the execution state of the AI ​​content generation thread, including:

[0083] Obtaining the execution progress information of the AI ​​content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, thereby determining the AI ​​content generation subtask currently executed by the user terminal; based on the currently executed AI content generation subtask, calling the corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask;

[0084] The execution speed of the AI ​​content generation thread is compared with a preset execution speed threshold. If the execution speed is less than the preset execution speed threshold, the corresponding processing component in the multimodal component set is updated to increase the execution speed of the AI ​​content generation thread.

[0085] The beneficial effect of the above embodiment is to obtain the execution progress information of the AI ​​content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, so as to determine the AI ​​content generation subtask currently executed by the user terminal, and call the corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask, so as to accurately determine the AI ​​content generation situation inside the user terminal. The execution speed of the AI ​​content generation thread is also compared with the preset execution speed threshold. If the execution speed is less than the preset execution speed threshold, the corresponding processing component in the multimodal component set is updated to increase the execution speed of the AI ​​content generation thread and improve the reliability of AI content generation at the user terminal.

[0086] In another embodiment, according to the relationship between the execution speed and a preset execution speed threshold, performing an abnormality determination on the operation of the AI ​​content generation thread includes:

[0087] If the execution speed is not less than the preset execution speed threshold, monitor the first operating parameter of the AI ​​content generation thread in real time, wherein the first operating parameter includes the amount of content data generated per unit time, memory utilization, the data amount difference between the input content and the output content, and the content generation bit error rate;

[0088] Obtaining a first operation evaluation coefficient using the content data volume generated per unit time, memory utilization, data volume difference between input content and output content, and content generation bit error rate included in the first operation parameter;

[0089] The first operation evaluation coefficient is obtained by the following formula:

[0090]

[0091] Among them, R 01 represents the first running evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C irepresents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; M i represents the memory utilization corresponding to the i-th unit time; E i represents the content generation bit error rate corresponding to the i-th unit time;

[0092] comparing the first operation evaluation coefficient with a preset coefficient threshold;

[0093] When the first operation evaluation coefficient exceeds a preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued;

[0094] If the execution speed is less than the preset execution speed threshold, the second operation parameter of the AI ​​content generation thread is monitored in real time, wherein the second operation parameter includes the amount of content data generated per unit time, the difference in data volume between the input content and the output content, and the execution speed value;

[0095] Obtaining a second operation evaluation coefficient using the content data volume generated per unit time, the data volume difference between the input content and the output content, and the execution speed value included in the second operation parameter;

[0096] The second operation evaluation coefficient is obtained by the following formula:

[0097]

[0098] Among them, R 02 represents the second operation evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; C zmaxi represents the maximum value of the data volume difference between the input content and the output content corresponding to the i-th unit time; v i represents the execution speed value generated corresponding to the i-th unit time; v b Indicates the guaranteed error of the generated execution speed value corresponding to n unit time;

[0099] comparing the second operation evaluation coefficient with a preset coefficient threshold;

[0100] When the second operation evaluation coefficient exceeds the preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued.

[0101] The beneficial effect of the above embodiment is that the scheme can dynamically select different operating parameters to monitor according to the relationship between the execution speed (i.e., processing efficiency) of the AI ​​content generation thread and the preset execution speed threshold. When the execution speed is not lower than the preset threshold, more comprehensive performance indicators (such as content data volume, memory utilization, data difference, bit error rate) are focused on; and when the execution speed is lower than the threshold, the focus is on generation efficiency and its related parameters (such as content data volume, data difference, execution speed). This dynamic adjustment improves the pertinence and efficiency of monitoring. By introducing the first operation evaluation coefficient (R01) and the second operation evaluation coefficient (R02), the scheme can comprehensively evaluate the operation status of the AI ​​content generation thread by combining multiple key operation parameters. This comprehensive evaluation method is more accurate than single parameter monitoring and can more comprehensively reflect the operation health of the thread. The change of the time series (n unit time) is taken into account in the coefficient calculation, which helps to capture the trend of performance fluctuations, thereby discovering potential operation anomalies in advance. When the operation evaluation coefficient exceeds the preset coefficient threshold, the system will immediately determine that the AI ​​content generation thread has an operation anomaly risk and trigger the early warning mechanism. This real-time anomaly detection and early warning capability helps to respond to and solve problems quickly, avoiding adverse consequences such as performance degradation or system crashes. Through continuous monitoring and anomaly determination, the solution helps to promptly discover and solve problems such as performance bottlenecks or improper resource allocation in AI content generation threads. This can not only improve the overall operating efficiency of the system, but also optimize resource utilization and enhance the stability and reliability of the system. The above technical solution has a wide range of applicability. It can be easily extended to other types of thread or process monitoring, providing strong support for system performance management and optimization.

[0102] In summary, this technical solution achieves accurate monitoring and abnormal judgment of the running status of AI content generation threads through dynamic monitoring, comprehensive evaluation, real-time early warning and other means, providing an effective technical means to improve system performance, optimize resource management and ensure system stability.

[0103] In another embodiment, the final generated content of the AI ​​content generation task is detected to determine whether the AI ​​content generation task is correctly executed; when the AI ​​content generation task is correctly executed, the final generated content is output, and the multimodal component set currently loaded by the user terminal is integrated and saved, including:

[0104] The final generated content of the AI ​​content generation task is detected to obtain a data bit error rate of the final generated content; if the data bit error rate is greater than or equal to a preset bit error rate threshold, it is determined that the AI ​​content generation task is not correctly executed; otherwise, it is determined that the AI ​​content generation task is correctly executed;

[0105] When the AI ​​content generation task is executed correctly, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user end is integrated into a repeatedly callable multimodal component set and saved in a designated storage space inside the user end.

[0106] The beneficial effect of the above embodiment is that the final generated content of the AI ​​content generation task is detected to obtain the data error rate of the final generated content, so as to accurately determine whether the AI ​​content generation task of the user end is correctly executed; in addition, when the AI ​​content generation task is correctly executed, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user end is integrated into a repeatable multimodal component set and saved in the designated storage space inside the user end, so as to ensure the normal display of the final generated content, facilitate the user end to reuse the multimodal component set for AIGC service in the future, reduce the complexity of AI content generation, and realize one-stop AIGC content generation.

[0107] See also Figure 2 As shown, a multimodal AIGC generation service control system provided by an embodiment of the present application includes:

[0108] A multimodal component determination module, configured to select processing components for different types of content from the AIGC library based on a content generation request from a user terminal, and to form a multimodal component set from all selected processing components;

[0109] A multimodal component matching module, used to load the multimodal component set into the user terminal based on the real-time working status information of the user terminal, and perform a multimodal component thread matching operation on the user terminal;

[0110] The AI ​​content generation task processing module is used to obtain the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, update the corresponding processing component in the multimodal component set to adjust the execution status of the AI ​​content generation thread;

[0111] An AI content generation task judgment module is used to detect the final generated content of the AI ​​content generation task and determine whether the AI ​​content generation task is executed correctly;

[0112] The multimodal component set operation module is used to output the final generated content when the AI ​​content generation task is executed correctly, and to integrate and save the multimodal component set currently loaded on the user end.

[0113] The beneficial effects of the above embodiments are as follows: the multimodal AIGC generation service control system selects processing components for different types of content from the AIGC library, and forms a multimodal component set with all the selected processing components, and loads the multimodal component set to the user end based on the real-time working status information of the user end, so that the user end can perform multimodal AIGC services; based on the execution status information of the AI ​​content generation task of the user end, the corresponding processing component is called from the multimodal component set to complete the corresponding AI content generation thread, so that different modal processing components can be used flexibly and quickly in the AI ​​content generation process, thereby improving the reliability and efficiency of AI content generation; and when the AI ​​content generation task is correctly executed, the final generated content of the AI ​​content generation task is output, and the multimodal component set currently loaded by the user end is integrated and saved, so that the user end can reuse the multimodal component set for AIGC services in the future, reduce the complexity of AI content generation, and realize one-stop AIGC content generation.

[0114] In another embodiment, the multimodal component determination module is used to select processing components for different types of content from the AIGC library based on a content generation request from a user terminal, and form a multimodal component set with all selected processing components, including:

[0115] Parse and process the content generation request from the user end to determine the modal attribute information of the AI ​​content that the user end currently expects to generate; based on the modal attribute information, select processing components for at least two of the text modality, image modality, video modality, and voice modality from the AIGC library, and form all the selected processing components into a multimodal component set;

[0116] The multimodal component matching module is used to load the multimodal component set to the user terminal based on the real-time working status information of the user terminal, and perform a multimodal component thread matching operation on the user terminal, including:

[0117] Based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined; based on the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power.

[0118] The beneficial effect of the above embodiment is that when the user terminal needs to obtain AIGC services and generate corresponding AI content, the content generation request from the user terminal is first parsed and processed to determine the modal attribute information of the AI ​​content that the user terminal currently expects to generate. The modal attribute information may include but is not limited to text modal attribute information, image modal attribute information, video modal attribute information, and voice modal attribute information, so that the AI ​​content that the user terminal currently expects to generate can be comprehensively identified in terms of modality. Then, based on the modal attribute information, processing components for at least two of the text modality, image modality, video modality, and voice modality are selected from the AIGC library, and all the selected processing components are combined into a multimodal component set, so that the formed multimodal component set can provide matching multimodal AIGC services for the user terminal. In order to ensure the normal formation of the multimodal component set inside the user terminal, it is necessary to run the multimodal component set with the help of the threads in the user terminal. At this time, based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined, and then according to the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power, so that the multimodal component set can be fully triggered and run inside the user terminal.

[0119] In another embodiment, the AI ​​content generation task processing module is used to obtain the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, update the corresponding processing component in the multimodal component set to adjust the execution status of the AI ​​content generation thread, including:

[0120] Obtaining the execution progress information of the AI ​​content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, thereby determining the AI ​​content generation subtask currently executed by the user terminal; based on the currently executed AI content generation subtask, calling the corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask;

[0121] The execution speed of the AI ​​content generation thread is compared with a preset execution speed threshold. If the execution speed is less than the preset execution speed threshold, the corresponding processing component in the multimodal component set is updated to increase the execution speed of the AI ​​content generation thread.

[0122] The beneficial effect of the above embodiment is to obtain the execution progress information of the AI ​​content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, so as to determine the AI ​​content generation subtask currently executed by the user terminal, and call the corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask, so as to accurately determine the AI ​​content generation situation inside the user terminal. The execution speed of the AI ​​content generation thread is also compared with the preset execution speed threshold. If the execution speed is less than the preset execution speed threshold, the corresponding processing component in the multimodal component set is updated to increase the execution speed of the AI ​​content generation thread and improve the reliability of AI content generation at the user terminal.

[0123] In another embodiment, according to the relationship between the execution speed and a preset execution speed threshold, performing an abnormality determination on the operation of the AI ​​content generation thread includes:

[0124] If the execution speed is not less than the preset execution speed threshold, monitor the first operating parameter of the AI ​​content generation thread in real time, wherein the first operating parameter includes the amount of content data generated per unit time, memory utilization, the data amount difference between the input content and the output content, and the content generation bit error rate;

[0125] Obtaining a first operation evaluation coefficient using the content data volume generated per unit time, memory utilization, data volume difference between input content and output content, and content generation bit error rate included in the first operation parameter;

[0126] The first operation evaluation coefficient is obtained by the following formula:

[0127]

[0128] Among them, R 01 represents the first running evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; M i represents the memory utilization corresponding to the i-th unit time; E i represents the content generation bit error rate corresponding to the i-th unit time;

[0129] comparing the first operation evaluation coefficient with a preset coefficient threshold;

[0130] When the first operation evaluation coefficient exceeds a preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued;

[0131] If the execution speed is less than the preset execution speed threshold, the second operation parameter of the AI ​​content generation thread is monitored in real time, wherein the second operation parameter includes the amount of content data generated per unit time, the difference in data volume between the input content and the output content, and the execution speed value;

[0132] Obtaining a second operation evaluation coefficient using the content data volume generated per unit time, the data volume difference between the input content and the output content, and the execution speed value included in the second operation parameter;

[0133] The second operation evaluation coefficient is obtained by the following formula:

[0134]

[0135] Among them, R 02 represents the second operation evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; C zmaxi represents the maximum value of the data volume difference between the input content and the output content corresponding to the i-th unit time; v i represents the execution speed value generated corresponding to the i-th unit time; v b Indicates the guaranteed error of the generated execution speed value corresponding to n unit time;

[0136] comparing the second operation evaluation coefficient with a preset coefficient threshold;

[0137] When the second operation evaluation coefficient exceeds the preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued.

[0138] The beneficial effect of the above embodiment is that the scheme can dynamically select different operating parameters to monitor according to the relationship between the execution speed (i.e., processing efficiency) of the AI ​​content generation thread and the preset execution speed threshold. When the execution speed is not lower than the preset threshold, more comprehensive performance indicators (such as content data volume, memory utilization, data difference, bit error rate) are focused on; and when the execution speed is lower than the threshold, the focus is on generation efficiency and its related parameters (such as content data volume, data difference, execution speed). This dynamic adjustment improves the pertinence and efficiency of monitoring. By introducing the first operation evaluation coefficient (R01) and the second operation evaluation coefficient (R02), the scheme can comprehensively evaluate the operation status of the AI ​​content generation thread by combining multiple key operation parameters. This comprehensive evaluation method is more accurate than single parameter monitoring and can more comprehensively reflect the operation health of the thread. The change of the time series (n unit time) is taken into account in the coefficient calculation, which helps to capture the trend of performance fluctuations, thereby discovering potential operation anomalies in advance. When the operation evaluation coefficient exceeds the preset coefficient threshold, the system will immediately determine that the AI ​​content generation thread has an operation anomaly risk and trigger the early warning mechanism. This real-time anomaly detection and early warning capability helps to respond to and solve problems quickly, avoiding adverse consequences such as performance degradation or system crashes. Through continuous monitoring and anomaly determination, the solution helps to promptly discover and solve problems such as performance bottlenecks or improper resource allocation in AI content generation threads. This can not only improve the overall operating efficiency of the system, but also optimize resource utilization and enhance the stability and reliability of the system. The above technical solution has a wide range of applicability. It can be easily extended to other types of thread or process monitoring, providing strong support for system performance management and optimization.

[0139] In summary, this technical solution achieves accurate monitoring and abnormal judgment of the running status of AI content generation threads through dynamic monitoring, comprehensive evaluation, real-time early warning and other means, providing an effective technical means to improve system performance, optimize resource management and ensure system stability.

[0140] In another embodiment, the AI ​​content generation task judgment module is used to detect the final generated content of the AI ​​content generation task to determine whether the AI ​​content generation task is correctly executed, including:

[0141] The final generated content of the AI ​​content generation task is detected to obtain a data bit error rate of the final generated content; if the data bit error rate is greater than or equal to a preset bit error rate threshold, it is determined that the AI ​​content generation task is not correctly executed; otherwise, it is determined that the AI ​​content generation task is correctly executed;

[0142] The multimodal component set operation module is used to output the final generated content when the AI ​​content generation task is executed correctly, and to integrate and save the multimodal component set currently loaded by the user end, including:

[0143] When the AI ​​content generation task is executed correctly, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user end is integrated into a repeatedly callable multimodal component set and saved in a designated storage space inside the user end.

[0144] The beneficial effect of the above embodiment is that the final generated content of the AI ​​content generation task is detected to obtain the data error rate of the final generated content, so as to accurately determine whether the AI ​​content generation task of the user end is correctly executed; in addition, when the AI ​​content generation task is correctly executed, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user end is integrated into a repeatable multimodal component set and saved in the designated storage space inside the user end, so as to ensure the normal display of the final generated content, facilitate the user end to reuse the multimodal component set for AIGC service in the future, reduce the complexity of AI content generation, and realize one-stop AIGC content generation.

[0145] In general, the multimodal AIGC generation service control method and system select processing components for different types of content from the AIGC library, and form all the selected processing components into a multimodal component set, and based on the real-time working status information of the user end, load the multimodal component set to the user end, so that the user end can perform multimodal AIGC services; based on the execution status information of the AI ​​content generation task of the user end, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread, so that different modal processing components can be used flexibly and quickly in the AI ​​content generation process, thereby improving the reliability and efficiency of AI content generation; and when the AI ​​content generation task is correctly executed, the final generated content of the AI ​​content generation task is output, and the multimodal component set currently loaded on the user end is integrated and saved, so that the user end can reuse the multimodal component set for AIGC services in the future, reduce the complexity of AI content generation, and realize one-stop AIGC content generation.

[0146] The above is only a specific implementation of the present invention, and any other improvements made based on the concept of the present invention are considered to be within the protection scope of the present invention.

Claims

1. A multi-modal AIGC generation service control method, characterized in that: include: Based on a content generation request from a user terminal, processing components for different types of content are selected from the AIGC library, and all the selected processing components are combined into a multimodal component set; based on real-time working status information of the user terminal, the multimodal component set is loaded into the user terminal, and a multimodal component thread matching operation is performed on the user terminal; Obtaining execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, calling a corresponding processing component from the multimodal component set to complete a corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, updating a corresponding processing component in the multimodal component set, thereby adjusting the execution status of the AI ​​content generation thread; The final generated content of the AI ​​content generation task is detected to determine whether the AI ​​content generation task is executed correctly; when the AI ​​content generation task is executed correctly, the final generated content is output, and the multimodal component set currently loaded on the user end is integrated and saved.

2. The multi-modal AIGC generation service control method according to claim 1, characterized in that: Based on the content generation request from the user end, processing components for different types of content are selected from the AIGC library, and all the selected processing components are combined into a multimodal component set; Based on the real-time working status information of the user terminal, the multimodal component set is loaded into the user terminal, and a multimodal component thread matching operation is performed on the user terminal, including: Parse and process the content generation request from the user end to determine the modal attribute information of the AI ​​content that the user end currently expects to generate; based on the modal attribute information, select processing components for at least two of the text modality, image modality, video modality, and voice modality from the AIGC library, and form all the selected processing components into a multimodal component set; Based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined; based on the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power.

3. The multi-modal AIGC generation service control method according to claim 1, characterized in that: Acquiring the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, calling the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, updating the corresponding processing component in the multimodal component set, thereby adjusting the execution status of the AI ​​content generation thread, including: Obtaining execution progress information of an AI content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, thereby determining the AI ​​content generation subtask currently executed by the user terminal; based on the currently executed AI content generation subtask, calling a corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask; comparing the execution speed of the AI ​​content generation thread with a preset execution speed threshold, and if the execution speed is less than the preset execution speed threshold, updating the corresponding processing component in the multimodal component set, thereby increasing the execution speed of the AI ​​content generation thread.

4. The multi-modal AIGC generation service control method according to claim 3, characterized in that: According to the relationship between the execution speed and the preset execution speed threshold, abnormality determination is performed on the operation of the AI ​​content generation thread, including: If the execution speed is not less than the preset execution speed threshold, the first operating parameter of the AI ​​content generation thread is monitored in real time, wherein the first operating parameter includes the amount of content data generated per unit time, memory utilization, the data amount difference between the input content and the output content, and the content generation bit error rate; Obtaining a first operation evaluation coefficient using the content data volume generated per unit time, memory utilization, data volume difference between input content and output content, and content generation bit error rate included in the first operation parameter; The first operation evaluation coefficient is obtained by the following formula: Among them, R 01 represents the first running evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; M i represents the memory utilization corresponding to the i-th unit time; E i represents the content generation bit error rate corresponding to the i-th unit time; comparing the first operation evaluation coefficient with a preset coefficient threshold; When the first operation evaluation coefficient exceeds a preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued; If the execution speed is less than the preset execution speed threshold, the second operation parameter of the AI ​​content generation thread is monitored in real time, wherein the second operation parameter includes the amount of content data generated per unit time, the difference in data volume between the input content and the output content, and the execution speed value; The second operation evaluation coefficient is obtained by using the content data volume generated per unit time, the data volume difference between the input content and the output content, and the execution speed value included in the second operation parameter; wherein the second operation evaluation coefficient is obtained by the following formula: Among them, R 02 represents the second operation evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; C zmaxi represents the maximum value of the data volume difference between the input content and the output content corresponding to the i-th unit time; v i Indicates the generated execution speed value corresponding to the i-th unit time; v b Indicates the guaranteed error of the generated execution speed value corresponding to n unit time; comparing the second operation evaluation coefficient with a preset coefficient threshold; When the second operation evaluation coefficient exceeds the preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued.

5. The multi-modal AIGC generation service control method according to claim 1, characterized in that: The final generated content of the AI ​​content generation task is detected to determine whether the AI ​​content generation task is correctly executed; when the AI ​​content generation task is correctly executed, the final generated content is output, and the multimodal component set currently loaded by the user terminal is integrated and saved, including: Detecting the final generated content of the AI ​​content generation task to obtain a data bit error rate of the final generated content; if the data bit error rate is greater than or equal to a preset bit error rate threshold, determining that the AI ​​content generation task is not correctly executed; otherwise, determining that the AI ​​content generation task is correctly executed; When the AI ​​content generation task is executed correctly, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user terminal is integrated into a repeatedly callable multimodal component set and saved in a designated storage space inside the user terminal.

6. A multi-modal AIGC generation service control system, characterized in that: include: A multimodal component determination module, configured to select processing components for different types of content from the AIGC library based on a content generation request from a user terminal, and to form a multimodal component set from all selected processing components; A multimodal component matching module, used to load the multimodal component set into the user terminal based on the real-time working status information of the user terminal, and perform a multimodal component thread matching operation on the user terminal; The AI ​​content generation task processing module is used to obtain the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, update the corresponding processing component in the multimodal component set to adjust the execution status of the AI ​​content generation thread; An AI content generation task judgment module, used to detect the final generated content of the AI ​​content generation task and determine whether the AI ​​content generation task is executed correctly; The multimodal component set operation module is used to output the final generated content when the AI ​​content generation task is executed correctly, and to integrate and save the multimodal component set currently loaded on the user end.

7. The multi-modal AIGC generation service control system according to claim 5, characterized in that: The multimodal component determination module is used to select processing components for different types of content from the AIGC library based on a content generation request from a user terminal, and form a multimodal component set with all the selected processing components, including: Parse and process the content generation request from the user end to determine the modal attribute information of the AI ​​content that the user end currently expects to generate; based on the modal attribute information, select processing components for at least two of the text modality, image modality, video modality, and voice modality from the AIGC library, and form all the selected processing components into a multimodal component set; The multimodal component matching module is used to load the multimodal component set to the user terminal based on the real-time working status information of the user terminal, and perform a multimodal component thread matching operation on the user terminal, including: Based on the real-time working status information of all threads under the user terminal, the thread with the largest idle computing power in the user terminal is determined; based on the position information of the thread with the largest idle computing power on the user terminal, the multimodal component set is loaded into the user terminal, and all processing components under the multimodal component set are matched to the corresponding links of the thread with the largest idle computing power.

8. The multi-modal AIGC generation service control system according to claim 6, characterized in that: The AI ​​content generation task processing module is used to obtain the execution status information of the AI ​​content generation task of the user terminal, and based on the execution status information, call the corresponding processing component from the multimodal component set to complete the corresponding AI content generation thread; based on the execution status information of the AI ​​content generation thread, update the corresponding processing component in the multimodal component set to adjust the execution status of the AI ​​content generation thread, including: Obtaining execution progress information of an AI content generation subtask queue corresponding to the AI ​​content generation task of the user terminal, thereby determining the AI ​​content generation subtask currently executed by the user terminal; based on the currently executed AI content generation subtask, calling a corresponding processing component from the multimodal component set to complete the AI ​​content generation thread corresponding to the AI ​​content generation subtask; comparing the execution speed of the AI ​​content generation thread with a preset execution speed threshold, and if the execution speed is less than the preset execution speed threshold, updating the corresponding processing component in the multimodal component set, thereby increasing the execution speed of the AI ​​content generation thread. According to the relationship between the execution speed and the preset execution speed threshold, abnormality determination is performed on the operation of the AI ​​content generation thread, including: If the execution speed is not less than the preset execution speed threshold, monitor the first operating parameter of the AI ​​content generation thread in real time, wherein the first operating parameter includes the amount of content data generated per unit time, memory utilization, the data amount difference between the input content and the output content, and the content generation bit error rate; Obtaining a first operation evaluation coefficient using the content data volume generated per unit time, memory utilization, data volume difference between input content and output content, and content generation bit error rate included in the first operation parameter; The first operation evaluation coefficient is obtained by the following formula: Among them, R 01 represents the first running evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; M i represents the memory utilization corresponding to the i-th unit time; E i represents the content generation bit error rate corresponding to the i-th unit time; comparing the first operation evaluation coefficient with a preset coefficient threshold; When the first operation evaluation coefficient exceeds a preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued; If the execution speed is less than the preset execution speed threshold, the second operation parameter of the AI ​​content generation thread is monitored in real time, wherein the second operation parameter includes the amount of content data generated per unit time, the difference in data volume between the input content and the output content, and the execution speed value; The second operation evaluation coefficient is obtained by using the content data volume generated per unit time, the data volume difference between the input content and the output content, and the execution speed value included in the second operation parameter; wherein the second operation evaluation coefficient is obtained by the following formula: Among them, R 02 represents the second operation evaluation coefficient; n represents the number of unit time experienced by the AI ​​content generation thread; C i represents the amount of content data generated corresponding to the i-th unit time; C zpi represents the average value of the data volume difference between the input content and the output content corresponding to the i-th unit time; C zmaxi represents the maximum value of the data volume difference between the input content and the output content corresponding to the i-th unit time; v i Indicates the generated execution speed value corresponding to the i-th unit time; v b Indicates the guaranteed error of the generated execution speed value corresponding to n unit time; comparing the second operation evaluation coefficient with a preset coefficient threshold; When the second operation evaluation coefficient exceeds the preset coefficient threshold, it is determined that the AI ​​content generation thread has an operation abnormality risk, and an abnormal risk warning is issued.

9. The multi-modal AIGC generation service control system according to claim 5, characterized in that: The AI ​​content generation task judgment module is used to detect the final generated content of the AI ​​content generation task to determine whether the AI ​​content generation task is correctly executed, including: Detecting the final generated content of the AI ​​content generation task to obtain a data bit error rate of the final generated content; if the data bit error rate is greater than or equal to a preset bit error rate threshold, determining that the AI ​​content generation task is not correctly executed; otherwise, determining that the AI ​​content generation task is correctly executed; The multimodal component set operation module is used to output the final generated content when the AI ​​content generation task is correctly executed, and to integrate and save the multimodal component set currently loaded by the user end, including: When the AI ​​content generation task is executed correctly, the final generated content is backed up and then output, and the multimodal component set currently loaded on the user terminal is integrated into a repeatedly callable multimodal component set and saved in a designated storage space inside the user terminal.

Images