AI order dispatching intelligent management and remote resource scheduling project management SaaS system oriented to multi-workshop industrial cooperation

Abstract

The invention discloses an AI order dispatching intelligent management and remote resource scheduling project management SaaS system oriented to multi-workshop industrial collaboration, and relates to the technical field of industrial data analysis. In the system, a heterogeneous data fusion module breaks a data island and constructs a unified data set; the data quality evaluation module quantitatively calculates a data credibility index Kxds to provide a reliable basis for decision making; a self-adaptive scheduling decision module dynamically switches a high-precision and robustness scheduling model according to Kxds to avoid the risk of model failure; the scheduling scheme execution and feedback module constructs a decision execution closed loop; and the system efficiency comprehensive evaluation module finally fits a global scheduling optimization index Yxzs by calculating model efficiency factors Gjyz and Ljyz and a matching degree index Ppds, and provides an accurate efficiency evaluation and continuous optimization basis for the system.

Description

Technical Field

[0001] This invention relates to the field of industrial data analysis technology, specifically to an AI-powered intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration. Background Technology

[0002] The origins of intelligent order dispatch management and remote resource scheduling project management SaaS systems can be traced back to manual production scheduling in the early stages of industrialization, later evolving into information management through Enterprise Resource Planning (ERP) and Manufacturing Execution Systems (MES). With the maturity of cloud computing and the SaaS model, enterprises can subscribe and use these systems without on-premises deployment, significantly lowering the application threshold. In recent years, the deep integration of artificial intelligence, big data, and the Internet of Things (IoT) technologies has driven the development of systems towards intelligence and predictive capabilities, enabling dynamic and optimal resource allocation.

[0003] However, in multi-workshop collaboration scenarios, data silos and heterogeneity issues often arise. Different workshops may possess equipment (such as PLCs and sensors) from different suppliers and from different eras, as well as independent management software (such as MES and WMS). The data they generate varies significantly in format, protocol, semantics, and timestamp accuracy. Although SaaS platforms provide a unified interface in the cloud, achieving low-cost, high-reliability, and real-time acquisition and cleaning of these massive amounts of heterogeneous data at the physical and data link layers, and standardizing it into a unified data model, remains a huge engineering challenge. This makes it impossible to obtain a global, accurate, and real-time view of production. The core intelligent scheduling algorithm of a system (whether based on operations research or machine learning) is typically built and trained on an idealized, standardized data model. However, due to the aforementioned data integration challenges, the input data acquired by the algorithm during actual operation is incomplete, delayed, or non-standard. This leads to "context failure"—the algorithm cannot understand the subtle but critical production rhythms, implicit constraints, or informal operating procedures of each workshop.

[0004] Furthermore, when minor anomalies occur in the data of a particular workshop (such as data retransmission causing timing discrepancies after a brief equipment outage), or when a new workshop brings in a new data paradigm, the global optimization model may make completely erroneous scheduling decisions that could even lead to production disruptions. It lacks the robustness and common-sense judgment of a human scheduler. In short, poor integration at the basic data level directly weakens the intelligence and reliability of higher-level algorithms, preventing them from delivering their theoretically optimal scheduling promises in complex, dynamic, multi-workshop real-world environments. Summary of the Invention

[0005] Technical problems to be solved In view of the deficiencies of the prior art, the application provides an AI order dispatching intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration, which solves the technical defects in the background art. Technical scheme

[0006] To achieve the above object, the application is implemented by the following technical scheme: an AI order dispatching intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration, comprising a heterogeneous data fusion module, a data quality evaluation module, a self-adaptive scheduling decision module, a scheduling scheme execution and feedback module and a system efficiency comprehensive evaluation module; The heterogeneous data fusion module is used for protocol conversion and data format unification of data sources of multiple heterogeneous workshops accessed to the system through a preset protocol adapter and a semantic model, so as to construct a global production data set; The data quality evaluation module is used for real-time monitoring of characteristics of each data stream in the global production data set, analysis of its performance in terms of integrity, timeliness and consistency, so as to construct a data state set; The self-adaptive scheduling decision module is used for dynamic selection and application of a scheduling algorithm model matched with the current data quality according to the data state set, so as to generate an optimal order dispatching and resource scheduling scheme; The scheduling scheme execution and feedback module is used for issuing the scheduling scheme to each workshop execution unit and real-time collection of production execution data to form a closed-loop feedback; The system efficiency comprehensive evaluation module is used for correlation analysis of data quality and execution effect of the scheduling scheme, so as to comprehensively evaluate the global scheduling optimization level of the current SaaS system.

[0007] The heterogeneous data fusion module comprises a protocol adaptation unit and a semantic alignment unit; The protocol adaptation unit is used for providing a plug-and-play virtualization interface for different industrial protocols, and seamlessly accessing physical layer and link layer data streams from devices and management software of different suppliers and different years to a cloud platform; The semantic alignment unit is used for mapping and converting data from different sources with different naming rules and units into a unified, standardized data model and semantic label by using a preset industrial knowledge graph, so as to eliminate data islands and form a global production view for upper-layer application understanding.

[0008] The data quality evaluation module comprises a data characteristic acquisition unit and a quality index quantization unit; The data feature acquisition unit is used to monitor and record the data flow of each workshop in the global production data set in real time during each monitoring period to obtain relevant data quality feature information, including data integrity rate Lwz, data timeliness Lsxx and data consistency Lyzx. The quality index quantification unit is used to calculate the mean values ​​of the data integrity rate (Lwz), data timeliness (Lsxx), and data consistency (Lyzx) of each data stream within the monitoring period, based on the data integrity rate (Lwz), data timeliness (Lsxx), and data consistency (Lyzx) within each monitoring period, combined with a statistical mean calculation algorithm, to calculate the mean values ​​of the data integrity rate, timeliness, and consistency of each data stream within the monitoring period.

[0009] The data quality assessment module is further used to calculate the data credibility index Kxds, and the calculation process includes the following steps: The integrity weighted value is obtained by multiplying the average integrity rate of the data stream within the monitoring period by the first preset weighting coefficient. Calculate the reciprocal of the average timeliness of the data stream within the monitoring period, and multiply the reciprocal by the second preset weighting coefficient to obtain the timeliness weighted value; The consistency mean of the data stream within the monitoring period is multiplied by a third preset weighting coefficient to obtain the consistency weighted value; finally, the integrity weighted value, the timeliness weighted value and the consistency weighted value are summed, and the sum is the data credibility index Kxds.

[0010] The adaptive scheduling decision module includes a scheduling strategy selection unit; The scheduling strategy selection unit is used to pre-set a reliability threshold Y, and to determine the current data quality's ability to support advanced scheduling algorithms by comparing the reliability threshold Y with the data reliability index Kxds. The specific details are as follows: If the data credibility index Kxds exceeds the credibility threshold Y, it is determined that the current data stream is in a high-quality state. The system will call a high-precision optimization model, including deep reinforcement learning and mixed integer programming, to perform order dispatch and resource scheduling. If the data credibility index Kxds does not exceed the credibility threshold Y, the current data stream is judged to be in a low-quality or unreliable state. The system will automatically switch to a robust heuristic model, including rule-based scheduling and simple algorithms with greater safety margins, to avoid risks caused by inaccurate data.

[0011] The adaptive scheduling decision module also includes a model switching execution unit; The model switching execution unit is used to perform corresponding operations after the scheduling strategy selection unit makes a switching decision. If it is determined that a switch to a robust heuristic model is required, the system will automatically increase the planned working time buffer, relax the equipment coupling constraints, and issue an alarm notification to the management personnel through the system interface or API, stating "Data quality has been degraded and conservative scheduling mode has been enabled," prompting them to pay attention to the data source status of the corresponding workshop.

[0012] The system performance comprehensive evaluation module includes a scheduling performance analysis unit; The scheduling performance analysis unit is used to calculate the high-precision model performance factor Gjyz and the robust model performance factor Ljyz, respectively; wherein, the calculation process of the high-precision model performance factor Gjyz includes: The plan achievement rate within the high-precision model scheduling period is multiplied by the first efficiency weight coefficient to obtain the plan achievement weighted value; the resource utilization rate within the same period is multiplied by the second efficiency weight coefficient to obtain the resource utilization weighted value; finally, the plan achievement weighted value and the resource utilization weighted value are summed. The calculation process of the robustness model effectiveness factor Ljyz includes: Calculate the reciprocal of the production anomaly rate within the robustness model scheduling cycle, and multiply the reciprocal by the third efficiency weight coefficient to obtain the production stability weighted value; calculate the reciprocal of the scheduling delay rate within the cycle, and multiply the reciprocal by the fourth efficiency weight coefficient to obtain the scheduling timeliness weighted value; finally, sum the production stability weighted value and the scheduling timeliness weighted value.

[0013] The system performance comprehensive evaluation module also includes a model matching degree analysis unit; The model matching degree analysis unit is used to analyze and calculate the model data matching degree index Ppds. The calculation process includes the following steps: Calculate the arithmetic mean of the performance factors Gjyz of all high-precision models within the total evaluation period, and multiply this mean by the weight of the first application time proportion to obtain the weighted mean of high-precision model performance. Calculate the arithmetic mean of the effectiveness factors Ljyz of all robust models within the total evaluation period, and multiply this mean by the weight of the second application time proportion to obtain the weighted mean of robust model effectiveness. The sum of the weighted mean of the high-precision model performance and the weighted mean of the robust model performance is the model data matching index Ppds.

[0014] The system performance comprehensive evaluation module also includes a comprehensive analysis unit; The comprehensive analysis unit is used to fit and calculate the global scheduling optimization index Yxzs. The calculation process includes the following steps: Multiply the data credibility index Kxds by the first comprehensive weighting coefficient to obtain the data credibility weighted index; multiply the model data matching index Ppds by the second comprehensive weighting coefficient to obtain the model matching weighted index; sum the data credibility weighted index, the model matching weighted index, and the preset correction constant to obtain the global scheduling optimization index Yxzs.

[0015] The system performance comprehensive evaluation module also includes an evaluation and rating unit; The evaluation and grading unit is used to pre-set the system health threshold W, and compares and analyzes the global scheduling optimization index Yxzs with the health threshold W to comprehensively evaluate the overall performance level of the current system and generate a corresponding optimization and maintenance report, the specific content of which is as follows: If the global scheduling optimization index Yxzs > the health threshold W, it indicates that the current system is in a highly efficient optimization state. The report will affirm the effectiveness of the current data governance and model application, and suggest that more advanced optimization algorithms can be introduced. If the global scheduling optimization index Yxzs ≤ the health threshold W, it indicates that the current system is in a sub-healthy or unoptimized state. The report will automatically trace the source, locate the data credibility index Kxds or model data matching index Ppds that caused the low score, and specifically point out which workshop's data source quality is poor or which scheduling model performs worse than expected, thereby providing operation and maintenance personnel with accurate guidance and suggestions for system optimization and data governance.

[0016] Beneficial Effects: This invention provides an AI-powered intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration. It offers the following beneficial effects: (1) This AI-powered intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration, by setting up a heterogeneous data fusion module and utilizing its internal protocol adaptation unit and semantic alignment unit, successfully solves the problem of data silos caused by different protocols and formats due to the diverse sources of equipment and software in multi-workshop collaboration scenarios; the system can seamlessly access data streams of different industrial protocols and map heterogeneous data into a standardized global production data set through industrial knowledge graphs, thereby providing an unprecedented global, accurate and real-time production view for upper-layer applications, laying a solid and reliable data foundation for realizing true global intelligent scheduling; (2) This AI-powered intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration achieves dynamic adaptive selection of scheduling algorithms by real-time quantitative calculation of data credibility index Kxds and comparison with preset credibility threshold Y. When the data quality is high, the system automatically calls a high-precision optimization model to pursue the optimal solution. When the data quality is poor, it switches to a robust heuristic model to ensure production stability and safety, thus completely solving the problems of "context failure" and "model fragility" caused by the reliance on idealized data in traditional scheduling algorithms. In addition, the system uses a system performance comprehensive evaluation module to quantitatively calculate the high-precision model performance factor Gjyz, the robust model performance factor Ljyz, and the model data matching index Ppds. Finally, it fits the global scheduling optimization index Yxzs and compares it with the system health threshold W. This not only ensures the reliability and efficiency of scheduling decisions in a dynamically changing data environment, but also provides a complete, data-driven closed-loop evaluation and optimization mechanism, significantly improving the intelligence level and industrial applicability of the entire system. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the system framework structure of the present invention; Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Please refer to Figure 1 This invention provides an AI-powered intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration, including a heterogeneous data fusion module, a data quality assessment module, an adaptive scheduling decision module, a scheduling scheme execution and feedback module, and a system performance comprehensive evaluation module. The heterogeneous data fusion module is used to convert protocols and unify data formats of data sources from multiple heterogeneous workshops connected to the system through a preset protocol adapter and semantic model, so as to construct a global production data set. The data quality assessment module is used to monitor the characteristics of each data stream in the global production data set in real time, and analyze its performance in terms of completeness, timeliness and consistency, so as to construct a data status set. The adaptive scheduling decision module is used to dynamically select and apply a scheduling algorithm model that matches the current data quality based on the data state set, so as to generate the optimal order dispatch and resource scheduling scheme. The scheduling scheme execution and feedback module is used to distribute the scheduling scheme to each workshop execution unit and collect production execution data in real time to form a closed-loop feedback. The system performance comprehensive evaluation module is used to correlate and analyze the data quality and the execution effect of the scheduling scheme in order to comprehensively evaluate the global scheduling optimization level of the current SaaS system.

[0020] In this embodiment, the heterogeneous data fusion module effectively breaks down data silos across multiple workshops, constructing a unified global production data set and providing a complete and standardized data foundation for global optimization. Based on this, the data quality assessment module analyzes data characteristics in real time and constructs a data state set, providing a quantified data reliability index Kxds for subsequent decision-making, ensuring the reliability of the decision input source. Furthermore, the adaptive scheduling decision module, based on the comparison between the data reliability index Kxds and a preset reliability threshold Y, dynamically and intelligently switches the scheduling algorithm, ensuring that the scheduling scheme can pursue optimal performance when data quality is high and maintain stability when data quality is low, effectively regulating... This avoids the risk of model failure. Subsequently, the scheduling scheme execution and feedback module accurately distributes the optimized scheme and collects real-time feedback, constructing a closed-loop management system from decision-making to execution and data feedback. Finally, the system performance comprehensive evaluation module correlates data quality and execution effect, and quantitatively calculates the high-precision model performance factor Gjyz, the robustness model performance factor Ljyz, and the model data matching index Ppds. Finally, it fits the global scheduling optimization index Yxzs and compares it with the system health threshold W, providing accurate data support and decision-making basis for the continuous optimization and performance evaluation of the system, significantly improving the intelligence, robustness, and practical value of the entire SaaS system.

[0021] Example 2: The heterogeneous data fusion module includes a protocol adaptation unit and a semantic alignment unit; The protocol adaptation unit is used to provide plug-and-play virtualization interfaces for different industrial protocols, and to seamlessly connect physical layer and link layer data streams from devices and management software from different vendors and eras to the cloud platform. The semantic alignment unit is used to map and convert data from different sources with different naming rules and units into a unified and standardized data model and semantic tags using a pre-built industrial knowledge graph, so as to eliminate data silos and form a global production view that can be understood by upper-layer applications.

[0022] The data quality assessment module includes a data feature acquisition unit and a quality indicator quantification unit; The data feature acquisition unit is used to monitor and record the data flow of each workshop in the global production data set in real time during each monitoring period to obtain relevant data quality feature information, including data integrity rate Lwz, data timeliness Lsxx and data consistency Lyzx. The quality index quantification unit is used to calculate the mean values ​​of the data integrity rate (Lwz), data timeliness (Lsxx), and data consistency (Lyzx) of each data stream within the monitoring period, based on the data integrity rate (Lwz), data timeliness (Lsxx), and data consistency (Lyzx) within each monitoring period, combined with a statistical mean calculation algorithm, to calculate the mean values ​​of the data integrity rate, timeliness, and consistency of each data stream within the monitoring period.

[0023] The data quality assessment module is further used to calculate the data credibility index Kxds, and the calculation process includes the following steps: The integrity weighted value is obtained by multiplying the average integrity rate of the data stream within the monitoring period by the first preset weighting coefficient. Calculate the reciprocal of the average timeliness of the data stream within the monitoring period, and multiply the reciprocal by the second preset weighting coefficient to obtain the timeliness weighted value; The consistency mean of the data stream within the monitoring period is multiplied by a third preset weighting coefficient to obtain the consistency weighted value; finally, the integrity weighted value, the timeliness weighted value and the consistency weighted value are summed, and the sum is the data credibility index Kxds.

[0024] The adaptive scheduling decision module includes a scheduling strategy selection unit; The scheduling strategy selection unit is used to pre-set a reliability threshold Y, and to determine the current data quality's ability to support advanced scheduling algorithms by comparing the reliability threshold Y with the data reliability index Kxds. The specific details are as follows: If the data credibility index Kxds exceeds the credibility threshold Y, it is determined that the current data stream is in a high-quality state. The system will call a high-precision optimization model, including deep reinforcement learning and mixed integer programming, to perform order dispatch and resource scheduling. If the data credibility index Kxds does not exceed the credibility threshold Y, the current data stream is judged to be in a low-quality or unreliable state. The system will automatically switch to a robust heuristic model, including rule-based scheduling and simple algorithms with greater safety margins, to avoid risks caused by inaccurate data.

[0025] The adaptive scheduling decision module also includes a model switching execution unit; The model switching execution unit is used to perform corresponding operations after the scheduling strategy selection unit makes a switching decision. If it is determined that a switch to a robust heuristic model is required, the system will automatically increase the planned working time buffer, relax the equipment coupling constraints, and issue an alarm notification to the management personnel through the system interface or API, stating "Data quality has been degraded and conservative scheduling mode has been enabled," prompting them to pay attention to the data source status of the corresponding workshop.

[0026] The system performance comprehensive evaluation module includes a scheduling performance analysis unit; The scheduling performance analysis unit is used to calculate the high-precision model performance factor Gjyz and the robust model performance factor Ljyz, respectively; wherein, the calculation process of the high-precision model performance factor Gjyz includes: The plan achievement rate within the high-precision model scheduling period is multiplied by the first efficiency weight coefficient to obtain the plan achievement weighted value; the resource utilization rate within the same period is multiplied by the second efficiency weight coefficient to obtain the resource utilization weighted value; finally, the plan achievement weighted value and the resource utilization weighted value are summed. The calculation process of the robustness model effectiveness factor Ljyz includes: Calculate the reciprocal of the production anomaly rate within the robustness model scheduling cycle, and multiply the reciprocal by the third efficiency weight coefficient to obtain the production stability weighted value; calculate the reciprocal of the scheduling delay rate within the cycle, and multiply the reciprocal by the fourth efficiency weight coefficient to obtain the scheduling timeliness weighted value; finally, sum the production stability weighted value and the scheduling timeliness weighted value.

[0027] The system performance comprehensive evaluation module also includes a model matching degree analysis unit; The model matching degree analysis unit is used to analyze and calculate the model data matching degree index Ppds. The calculation process includes the following steps: Calculate the arithmetic mean of the performance factors Gjyz of all high-precision models within the total evaluation period, and multiply this mean by the weight of the first application time proportion to obtain the weighted mean of high-precision model performance. Calculate the arithmetic mean of the effectiveness factors Ljyz of all robust models within the total evaluation period, and multiply this mean by the weight of the second application time proportion to obtain the weighted mean of robust model effectiveness. The sum of the weighted mean of the high-precision model performance and the weighted mean of the robust model performance is the model data matching index Ppds.

[0028] The system performance comprehensive evaluation module also includes a comprehensive analysis unit; The comprehensive analysis unit is used to fit and calculate the global scheduling optimization index Yxzs. The calculation process includes the following steps: Multiply the data credibility index Kxds by the first comprehensive weighting coefficient to obtain the data credibility weighted index; multiply the model data matching index Ppds by the second comprehensive weighting coefficient to obtain the model matching weighted index; sum the data credibility weighted index, the model matching weighted index, and the preset correction constant to obtain the global scheduling optimization index Yxzs.

[0029] The system performance comprehensive evaluation module also includes an evaluation and rating unit; The evaluation and grading unit is used to pre-set the system health threshold W, and compares and analyzes the global scheduling optimization index Yxzs with the health threshold W to comprehensively evaluate the overall performance level of the current system and generate a corresponding optimization and maintenance report, the specific content of which is as follows: If the global scheduling optimization index Yxzs > the health threshold W, it indicates that the current system is in a highly efficient optimization state. The report will affirm the effectiveness of the current data governance and model application, and suggest that more advanced optimization algorithms can be introduced. If the global scheduling optimization index Yxzs ≤ the health threshold W, it indicates that the current system is in a sub-healthy or unoptimized state. The report will automatically trace the source, locate the data credibility index Kxds or model data matching index Ppds that caused the low score, and specifically point out which workshop's data source quality is poor or which scheduling model performs worse than expected, thereby providing operation and maintenance personnel with accurate guidance and suggestions for system optimization and data governance.

[0030] In this embodiment, the challenges of underlying data access and understanding are first addressed through the protocol adaptation unit and semantic alignment unit, laying a solid foundation for upper-layer intelligence. Building upon this, the system collects data integrity rate (Lwz), data timeliness (Lsxx), and data consistency (Lyzx) to create a basic profile of the original data stream from three core dimensions: integrity, real-time performance, and logical consistency, providing a basis for quantitative evaluation. The weighted calculation of the data credibility index (Kxds) from these basic indicators abstracts multi-dimensional and fragmented data quality characteristics into a single, clear, and machine-determined trust score. Subsequently, Kxds is compared with a preset credibility threshold (Y). The core meaning of this comparison is to establish a dynamic decision-making trigger, enabling the system to make the most reasonable choice between pursuing efficiency (high-precision model) and ensuring stability (robust model) based on the quality of the data, thereby achieving the core... The system employs adaptive scheduling. To evaluate the long-term effectiveness of this decision-making mechanism, the system calculates the high-precision model performance factor Gjyz and the robust model performance factor Ljyz, respectively. This objectively measures the actual performance of the two different strategies in their respective application scenarios. Furthermore, the calculation of the model data matching index Ppds assesses the intelligence of the system's model selection, i.e., the matching quality between the model switching strategy and actual data quality changes. Finally, the comprehensive calculation of the global scheduling optimization index Yxzs and its comparison with the system health threshold W provides a top-level, quantifiable comprehensive health score for the entire SaaS system's intelligence level. This score can be traced back to specific data source problems (low Kxds) or model matching problems (low Ppds), thus providing operations and maintenance personnel with precise optimization directions and realizing a complete intelligent closed loop from data to decision-making to performance evaluation and continuous improvement.

[0031] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A SaaS system for AI-powered intelligent order dispatching and remote resource scheduling project management for multi-workshop industrial collaboration, characterized by: It includes a heterogeneous data fusion module, a data quality assessment module, an adaptive scheduling decision module, a scheduling scheme execution and feedback module, and a system performance comprehensive assessment module; The heterogeneous data fusion module is used to convert protocols and unify data formats of data sources from multiple heterogeneous workshops connected to the system through a preset protocol adapter and semantic model, so as to construct a global production data set. The data quality assessment module is used to monitor the characteristics of each data stream in the global production data set in real time, and analyze its performance in terms of completeness, timeliness and consistency, so as to construct a data status set. The adaptive scheduling decision module is used to dynamically select and apply a scheduling algorithm model that matches the current data quality based on the data state set, so as to generate the optimal order dispatch and resource scheduling scheme. The scheduling scheme execution and feedback module is used to distribute the scheduling scheme to each workshop execution unit and collect production execution data in real time to form a closed-loop feedback. The system performance comprehensive evaluation module is used to correlate and analyze the data quality and the execution effect of the scheduling scheme in order to comprehensively evaluate the global scheduling optimization level of the current SaaS system.

2. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The heterogeneous data fusion module includes a protocol adaptation unit and a semantic alignment unit; The protocol adaptation unit is used to provide plug-and-play virtualization interfaces for different industrial protocols, and to seamlessly connect physical layer and link layer data streams from devices and management software from different vendors and eras to the cloud platform. The semantic alignment unit is used to map and convert data from different sources with different naming rules and units into a unified and standardized data model and semantic tags using a pre-built industrial knowledge graph, so as to eliminate data silos and form a global production view that can be understood by upper-layer applications.

3. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The data quality assessment module includes a data feature acquisition unit and a quality indicator quantification unit; The data feature acquisition unit is used to monitor and record the data flow of each workshop in the global production data set in real time during each monitoring period to obtain relevant data quality feature information, including data integrity rate Lwz, data timeliness Lsxx and data consistency Lyzx. The quality index quantification unit is used to calculate the mean values ​​of the data integrity rate (Lwz), data timeliness (Lsxx), and data consistency (Lyzx) of each data stream within the monitoring period, based on the data integrity rate (Lwz), data timeliness (Lsxx), and data consistency (Lyzx) within each monitoring period, combined with a statistical mean calculation algorithm, to calculate the mean values ​​of the data integrity rate, timeliness, and consistency of each data stream within the monitoring period.

4. The AI-powered intelligent management and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The data quality assessment module is further used to calculate the data credibility index Kxds, and the calculation process includes the following steps: The integrity weighted value is obtained by multiplying the average integrity rate of the data stream within the monitoring period by the first preset weighting coefficient. Calculate the reciprocal of the average timeliness of the data stream within the monitoring period, and multiply the reciprocal by the second preset weighting coefficient to obtain the timeliness weighted value; The consistency mean of the data stream within the monitoring period is multiplied by a third preset weighting coefficient to obtain the consistency weighted value; finally, the integrity weighted value, the timeliness weighted value and the consistency weighted value are summed, and the sum is the data credibility index Kxds.

5. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The adaptive scheduling decision module includes a scheduling strategy selection unit; The scheduling strategy selection unit is used to pre-set a reliability threshold Y, and to determine the current data quality's ability to support advanced scheduling algorithms by comparing the reliability threshold Y with the data reliability index Kxds. The specific details are as follows: If the data credibility index Kxds exceeds the credibility threshold Y, it is determined that the current data stream is in a high-quality state. The system will call a high-precision optimization model, including deep reinforcement learning and mixed integer programming, to perform order dispatch and resource scheduling. If the data credibility index Kxds does not exceed the credibility threshold Y, the current data stream is judged to be in a low-quality or unreliable state. The system will automatically switch to a robust heuristic model, including rule-based scheduling and simple algorithms with greater safety margins, to avoid risks caused by inaccurate data.

6. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The adaptive scheduling decision module also includes a model switching execution unit; The model switching execution unit is used to perform corresponding operations after the scheduling strategy selection unit makes a switching decision. If it is determined that a switch to a robust heuristic model is required, the system will automatically increase the planned working time buffer, relax the equipment coupling constraints, and issue an alarm notification to the management personnel through the system interface or API, stating "Data quality has been degraded and conservative scheduling mode has been enabled," prompting them to pay attention to the data source status of the corresponding workshop.

7. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The system performance comprehensive evaluation module includes a scheduling performance analysis unit; The scheduling performance analysis unit is used to calculate the high-precision model performance factor Gjyz and the robust model performance factor Ljyz, respectively; wherein, the calculation process of the high-precision model performance factor Gjyz includes: The plan achievement rate within the high-precision model scheduling period is multiplied by the first efficiency weight coefficient to obtain the plan achievement weighted value; the resource utilization rate within the same period is multiplied by the second efficiency weight coefficient to obtain the resource utilization weighted value; finally, the plan achievement weighted value and the resource utilization weighted value are summed. The calculation process of the robustness model effectiveness factor Ljyz includes: Calculate the reciprocal of the production anomaly rate within the robustness model scheduling cycle, and multiply the reciprocal by the third efficiency weight coefficient to obtain the production stability weighted value; calculate the reciprocal of the scheduling delay rate within the cycle, and multiply the reciprocal by the fourth efficiency weight coefficient to obtain the scheduling timeliness weighted value; finally, sum the production stability weighted value and the scheduling timeliness weighted value.

8. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The system performance comprehensive evaluation module also includes a model matching degree analysis unit; The model matching degree analysis unit is used to analyze and calculate the model data matching degree index Ppds. The calculation process includes the following steps: Calculate the arithmetic mean of the performance factors Gjyz of all high-precision models within the total evaluation period, and multiply this mean by the weight of the first application time proportion to obtain the weighted mean of high-precision model performance. Calculate the arithmetic mean of the effectiveness factors Ljyz of all robust models within the total evaluation period, and multiply this mean by the weight of the second application time proportion to obtain the weighted mean of robust model effectiveness. The sum of the weighted mean of the high-precision model performance and the weighted mean of the robust model performance is the model data matching index Ppds.

9. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The system performance comprehensive evaluation module also includes a comprehensive analysis unit; The comprehensive analysis unit is used to fit and calculate the global scheduling optimization index Yxzs. The calculation process includes the following steps: Multiply the data credibility index Kxds by the first comprehensive weighting coefficient to obtain the data credibility weighted index; multiply the model data matching index Ppds by the second comprehensive weighting coefficient to obtain the model matching weighted index; sum the data credibility weighted index, the model matching weighted index, and the preset correction constant to obtain the global scheduling optimization index Yxzs.

10. The AI-powered intelligent order dispatching and remote resource scheduling project management SaaS system for multi-workshop industrial collaboration as described in claim 1, characterized in that: The system performance comprehensive evaluation module also includes an evaluation and rating unit; The evaluation and grading unit is used to pre-set the system health threshold W, and compares and analyzes the global scheduling optimization index Yxzs with the health threshold W to comprehensively evaluate the overall performance level of the current system and generate a corresponding optimization and maintenance report, the specific content of which is as follows: If the global scheduling optimization index Yxzs is greater than the health threshold W, it indicates that the current system is in a highly efficient optimization state. The report will affirm the effectiveness of the current data governance and model application, and suggest trying to introduce more advanced optimization algorithms. If the global scheduling optimization index Yxzs ≤ the health threshold W, it indicates that the current system is in a sub-healthy or unoptimized state. The report will automatically trace the source, locate the data credibility index Kxds or model data matching index Ppds that caused the low score, and specifically point out which workshop's data source quality is poor or which scheduling model performs worse than expected, thereby providing operation and maintenance personnel with accurate guidance and suggestions for system optimization and data governance.

Images