Manufacturing and service collaborative management prototype system based on supply chain integration

Abstract

The present application relates to the technical field of data processing, in particular to a manufacturing and service collaborative management and control prototype system based on supply chain integration, which comprises a data analysis module, a mutation detection module, a tensor rearrangement module, a collaborative management and control module and a redistribution module.In the present application, a unified state sequence is constructed by fusing the equipment running current vibration amplitude and order time sequence information, realizing the correlation expression of multi-source data in the same time dimension, identifying the running process mutation time combined with the adjacent state difference, capturing the production anomaly and task conflict, on this basis, the priority of the mutation and non-mutation period is rearranged, forming a data representation with time sequence weight, further combining the logistics delivery scale and task overlap relationship to aggregate the business connectivity, making the resource association more clear, at the same time, introducing the equipment state evaluation remaining time difference, realizing the synchronization of production rhythm and logistics capacity, and through parameter truncation and transport capacity redistribution, the scheduling decision has dynamic adaptation ability.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a prototype system for collaborative management and control of manufacturing and services based on supply chain integration. Background Technology

[0002] The field of data processing technology involves the collection, storage, transmission, calculation, analysis and application of data, covering core matters such as business data modeling, database structure design, data exchange mechanisms, process scheduling rules and cross-system information integration. It is mainly aimed at information organization and decision support in industrial, commercial and service scenarios, and realizes unified management and collaborative processing of production resource logistics information and service activities by establishing the relationship between data.

[0003] Among them, the traditional manufacturing and service collaborative management prototype system refers to the basic system form built around the information interaction and business collaboration among multiple participants in the supply chain. It addresses the technical issues of data inconsistency, discontinuous process connection, and scattered resource scheduling between the manufacturing and service links. The common approach includes entering procurement information, production plan, inventory quantity, and delivery node data into a relational database by pre-defining order field structures, using fixed-format data tables to transmit information between supplier production workshops and warehousing nodes, achieving process connection through production plan tables and logistics shipping lists with set sequences, adjusting production tasks according to manually set time nodes and inventory threshold rules, and synchronizing data between different systems through interface file exchange to achieve basic collaborative management of the manufacturing and service processes in the supply chain environment.

[0004] Existing technologies rely on predefined data fields and fixed table structures for information organization. Data updates depend on manual entry and periodic synchronization mechanisms, making it difficult to reflect the real-time status of multi-source operations. There are inconsistencies in time granularity between manufacturing and service processes. Process connections rely on sequential planning tables and manual threshold rules, lacking the ability to perceive dynamic fluctuations in the operation process. Resource scheduling is based on static inventory and time nodes, making it difficult to identify production anomalies and task conflicts. Data transfer across systems is achieved through interface file exchange, resulting in significant information delays and discontinuities. This further causes a disconnect between logistics arrangements and production rhythm, leading to sluggish responses in the overall collaboration process and affecting the accuracy and consistency of decision-making. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a prototype system for collaborative management and control of manufacturing and services based on supply chain integration. The technical solution is as follows: On the one hand, a prototype system for collaborative management and control of manufacturing and services based on supply chain integration was provided, which includes: The data analysis module acquires operating current, equipment vibration amplitude, order generation time, and service priority through sensors at the production workstations in the supply chain and order terminals. It normalizes the operating current and equipment vibration amplitude, aligns the order generation time and service priority time, generates a multi-source state sequence, and transmits it to the mutation detection module. The mutation detection module obtains the difference in current amplitude and vibration amplitude between adjacent times in the multi-source state sequence and calculates the feature distance. It then marks abnormal time points by combining the preset feature distance benchmark threshold, generates state mutation time, and transmits it to the tensor rearrangement module. The tensor rearrangement module extracts the mutation point and non-mutation point times based on the state mutation time, assigns priority values ​​to the mutation point times and rearranges the mutation point and non-mutation point times, generates rearranged tensor data and transmits it to the collaborative management module. The collaborative management module, based on the rearranged tensor data, obtains the total logistics delivery volume, task overlap duration, and equipment slice status. It performs business connectivity clustering on the total logistics delivery volume and task overlap duration and maps the spatial coordinates of business entities in the same cluster. It calculates the remaining time difference in combination with the equipment slice status and generates a collaborative management instruction set. The reallocation module obtains accelerated production compensation parameters and logistics capacity scheduling parameters based on the collaborative management and control instruction set, compares them with preset equipment load capacity thresholds, truncates the accelerated production compensation parameters, and simultaneously reallocates capacity from the collaborative management and control instruction set to generate a collaborative management and control reallocation instruction set.

[0006] As a further embodiment of the present invention, the multi-source state sequence includes current normalization, vibration normalization, and time alignment stamp; the state mutation time includes anomaly time index, mutation amplitude marker, and time location label; the rearranged tensor data includes mutation priority weight, non-mutation order index, and time rearrangement sequence; the collaborative management and control instruction set includes service connectivity cluster identifier, spatial mapping coordinates, remaining time difference, and equipment slice code; and the collaborative management and control reallocation instruction set includes compensation truncation parameters, capacity allocation ratio, and scheduling priority code.

[0007] As a further aspect of the present invention, the data analysis module includes: The signal analysis submodule acquires sensor signals from production workstations in the supply chain and order terminal records, collects operating current sequences and equipment vibration amplitude sequences, monitors order generation timestamps and service priority identifier sequences, and performs time stamp matching on multiple types of data according to a unified time scale to obtain multi-source raw data sequences. The normalization processing submodule calls the running current sequence and the equipment vibration amplitude sequence based on the multi-source original data sequence, performs mean normalization calculation on the running current sequence and the equipment vibration amplitude sequence, and aligns and splices the order generation time and service priority on the time axis to obtain the normalized feature sequence. The sequence construction submodule extracts normalized operating current data and normalized vibration amplitude data based on the normalized feature sequence, and calls the order generation time and service priority identifier after time axis alignment to perform field-level splicing and sequential arrangement operations to generate a multi-source state sequence.

[0008] As a further aspect of the present invention, the mutation detection module includes: The amplitude difference construction submodule obtains the current amplitude difference and vibration amplitude difference between adjacent times based on the multi-source state sequence, obtains the current amplitude difference sequence by differentiating the current sampling under the continuous time index, obtains the vibration amplitude difference sequence by differentiating the vibration sampling at the corresponding time, and aligns it with the current amplitude difference sequence in time to obtain the amplitude difference combination vector. The distance measurement submodule, based on the amplitude difference combination vector, calls the difference and sum of squares operation between vector components to construct a distance calculation expression. It multiplies the current amplitude difference and vibration amplitude difference under the same time index by the corresponding weight coefficients to unify the dimensions, and then performs differential square accumulation and square root to obtain the feature distance sequence. The anomaly marking submodule obtains the distance at multiple time points based on the feature distance sequence and compares it point by point with a preset feature distance benchmark threshold. Time indices that exceed the feature distance benchmark threshold are marked as anomaly index sets. The anomaly index sets are then time-mapped to generate a state change time series.

[0009] As a further embodiment of the present invention, the feature distance reference threshold is obtained by acquiring the operating current and vibration amplitude of the equipment within the calibration period, extracting the difference in current amplitude and the difference in vibration amplitude between adjacent times, multiplying them by the corresponding weight coefficients, performing differential square accumulation and square root processing to generate a calibration distance sequence, calculating the arithmetic mean and standard deviation of the calibration distance sequence, and summing the arithmetic mean with a preset multiple of the standard deviation to determine the distance.

[0010] As a further aspect of the present invention, the tensor rearrangement module includes: The coordinate extraction submodule extracts the time coordinates of the mutation point and the non-mutation point based on the state mutation time and the multi-source state sequence, matches the state mutation time and locates the time in the multi-source state sequence, records the positions of the mutation point and the non-mutation point according to the time and normalizes the coordinates to obtain a set of time coordinates. The priority assignment submodule assigns priority numbers to the time coordinates of mutation points according to the time coordinate set, performs numerical mapping on the location indices of multiple mutation points and assigns incremental numbers according to the order of mutation occurrence, sorts the numbering results, and generates a priority number sequence. The index rearrangement submodule rearranges the time coordinates of mutation points and non-mutation points based on the priority number sequence. It inserts the mutation point indexes into the beginning of the time series in numerical order, while arranging the non-mutation point indexes in their original order. It then performs index splicing and tensor structure recombination to obtain rearranged tensor data.

[0011] As a further aspect of the present invention, the collaborative management module includes: The logistics overlap submodule, based on the rearranged tensor data, obtains the business entity identifier and logistics path, detects the transmission records between multiple path points and accumulates them to obtain the total logistics transmission volume, calculates the time intersection length by combining the task time series and multiplies it by the duration conversion factor related to the total logistics transmission volume, and generates the entity overlap duration set. The connectivity clustering submodule, based on the entity overlap duration set, calls the product operation of the total logistics transfer volume and the intersection length to obtain the connectivity, sorts them in descending order according to the connectivity and divides them into multiple data clusters in equal quantities, extracts the entity spatial coordinate sequence within the cluster and performs centering adjustment to obtain the entity coordinate sequence of the same cluster. The instruction generation submodule obtains the device slice status and calculates the remaining duration, calls the coordinate sequence of the same cluster entity to match the associated slice number, performs difference calculation on the remaining duration of the slices in the same cluster to extract the remaining time difference sequence, and reassembles the instruction fields according to the slice number and the time difference sequence to generate a collaborative control instruction set.

[0012] As a further aspect of the present invention, the redistribution module includes: The parameter parsing submodule, based on the collaborative management and control instruction set, obtains accelerated production compensation parameters and logistics capacity scheduling parameters, aligns fields according to time tags and equipment identifiers, and performs index mapping and element concatenation of accelerated production compensation and logistics capacity scheduling according to equipment number to establish a joint parameter mapping set; The threshold verification submodule calls the parameter joint mapping set, calculates the difference between each accelerated production compensation parameter and the preset equipment rated load benchmark threshold, determines the sign attribute of the difference and marks the parameter items that exceed the equipment rated load benchmark threshold, records the corresponding parameter index and the over-limit boundary, and obtains the load over-limit boundary set. The capacity reconfiguration submodule extracts the over-limit parameters and corresponding boundaries from the load over-limit boundary set, performs boundary trimming on the over-limit parameters, splices and reassembles the trimmed accelerated production compensation parameters and logistics capacity scheduling parameters, and performs capacity allocation ratio multiplication calculation to generate a collaborative management and control redistribution instruction set.

[0013] As a further aspect of the present invention, the rated load reference threshold of the equipment is obtained by acquiring the operating power and rated current within the equipment calibration period, and using the operating power divided by the rated voltage to calculate the rated current, etc., in accordance with the laws of physics, to form a calibration load sequence. The arithmetic mean and standard deviation of the calibration load sequence are calculated, the standard deviation is multiplied by a preset multiple to obtain the deviation, and the arithmetic mean and the deviation are summed to determine the value.

[0014] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: By integrating equipment operating current vibration amplitude and order timing information to construct a unified state sequence, multi-source data can be correlated and expressed in the same time dimension. By combining the differences between adjacent states and feature distances to identify the moment of sudden change in the operation process, production anomalies and task conflicts can be captured in a timely manner. On this basis, the priority of sudden and non-sudden change periods is rearranged to form a structured data representation with time-series weights. Furthermore, by combining the scale of logistics delivery and the relationship of task overlap, business connectivity aggregation and spatial mapping are performed to make resource relationships clearer. At the same time, the remaining time difference of equipment slice status assessment is introduced to achieve synchronous measurement of production rhythm and logistics capacity. Through parameter truncation and capacity reallocation under constraints, scheduling decisions have dynamic adaptability and improve the response lag and deviation problems in the coordination process. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the system of the present invention; Figure 2 This is a schematic diagram of the system framework of the present invention; Figure 3 This is a flowchart of the data analysis module in this invention; Figure 4 This is a flowchart of the mutation detection module in this invention; Figure 5 This is a flowchart of the tensor rearrangement module in this invention; Figure 6 This is a flowchart of the collaborative management and control module in this invention; Figure 7 This is a flowchart of the redistribution module in this invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] This invention provides a prototype system for collaborative management and control of manufacturing and services based on supply chain integration, such as... Figure 1-2 The diagram shown illustrates a prototype system for collaborative management and control of manufacturing and services based on supply chain integration. This system includes: The data analysis module acquires operating current, equipment vibration amplitude, order generation time, and service priority through sensors at the production workstations in the supply chain and order terminals. It normalizes the operating current and equipment vibration amplitude, aligns the order generation time and service priority time, generates a multi-source state sequence, and transmits it to the mutation detection module. The mutation detection module obtains the difference in current amplitude and vibration amplitude between adjacent time points in the multi-source state sequence and calculates the feature distance. It then marks abnormal time points by combining the preset feature distance benchmark threshold, generates the state mutation time, and transmits it to the tensor rearrangement module. The tensor rearrangement module extracts the time of mutation points and non-mutation points based on the state mutation time, assigns priority values ​​to the time of mutation points and rearranges the time of mutation points and non-mutation points, generates rearranged tensor data and transmits it to the collaborative management module. The collaborative management module, based on rearranged tensor data, obtains the total logistics delivery volume, task overlap duration and equipment slice status. It performs business connectivity clustering on the total logistics delivery volume and task overlap duration and maps the spatial coordinates of business entities in the same cluster. It calculates the remaining time difference in combination with the equipment slice status and generates a collaborative management instruction set. The reallocation module obtains accelerated production compensation parameters and logistics capacity scheduling parameters based on the collaborative management and control instruction set, compares them with preset equipment load capacity thresholds, truncates the accelerated production compensation parameters, and simultaneously reallocates capacity using the collaborative management and control instruction set to generate a collaborative management and control reallocation instruction set.

[0020] The multi-source state sequence includes current normalization, vibration normalization, and time alignment stamp; the state mutation time includes anomaly time index, mutation amplitude marker, and time location label; the rearranged tensor data includes mutation priority weight, non-mutation order index, and time rearrangement sequence; the collaborative management and control instruction set includes business connectivity cluster identifier, spatial mapping coordinates, remaining time difference, and equipment slice encoding; and the collaborative management and control reallocation instruction set includes compensation truncation parameters, capacity allocation ratio, and scheduling priority encoding.

[0021] Specifically, such as Figure 2 , 3 As shown, the data analysis module includes: The signal analysis submodule acquires sensor signals from production workstations in the supply chain and order terminal records, collects operating current sequences and equipment vibration amplitude sequences, monitors order generation timestamps and service priority identifier sequences, and performs time stamp matching on multiple types of data according to a unified time scale to obtain multi-source raw data sequences. The signal data access unit is activated and a direct transmission link is established with the programmable logic controllers (PLCs) and enterprise resource planning (ERP) systems at the production workstations in the non-ferrous metallurgical supply chain. It reads signals from underlying sensors and records from top-level order terminals. The production sequence acquisition unit retrieves the operating current sequence from the three-phase operating data of the main motor at the workstation, with a fixed sampling frequency of 50 Hz and a single sequence length of 500 sampling points. Simultaneously, it retrieves the triaxial vibration amplitude sequence of the key bearing at the workstation, with a sampling frequency of 100 Hz. The business instruction extraction unit monitors the order generation timestamps issued by the ERP system. These timestamps are recorded in an absolute time format accurate to milliseconds. It also reads the corresponding service priority identifier sequence. For the expedited, regular, and deferred order types specified in the identifiers, the business instruction extraction unit quantifies expedited orders as a value of 3, regular orders as a value of 2, and deferred orders as a value of 1. The time axis alignment unit performs time stamp matching operations on the above multiple data types according to a unified time scale. The time stamp matching component performs alignment deviation calculation to determine the validity of the current data point. It acquires the device sampling timestamp and the service delivery timestamp, subtracts the device sampling timestamp from the service delivery timestamp to obtain the absolute value of the time stamp deviation, and determines whether this absolute value is within the time alignment tolerance threshold. If it is, it is considered an aligned data point. The time alignment tolerance threshold is obtained statistically from collecting 1000 historical network latency records. Experimental data shows that 99% of transmission latency is between 0 and 5 milliseconds. The time axis alignment unit strictly sets the time alignment tolerance threshold to 5 milliseconds. Based on actual operational data, when the time offset of the service delivery timestamp relative to the baseline zero point is 1005 milliseconds and the time offset of the device sampling timestamp relative to the baseline zero point is 1002 milliseconds, the absolute value of the time stamp deviation is 3 milliseconds. This deviation is less than the time alignment tolerance threshold of 5 milliseconds, meeting the data alignment baseline condition. The time axis alignment unit horizontally aggregates the corresponding operating current sequence value, device vibration amplitude sequence value, order generation timestamp, and service priority identifier value to obtain a multi-source original data sequence.

[0022] Table 1. Alignment Results of Multi-Source Original Data Sequences 1 45 2.5 1005 3 2 44 2.4 1025 3 Table 1 shows the multi-source raw data sequence fragment records after time alignment processing, which clearly lists the correspondence between the values ​​of each sensor and the quantization dimension.

[0023] The normalization processing submodule calls the running current sequence and equipment vibration amplitude sequence based on the multi-source raw data sequence, performs mean normalization calculation on the running current sequence and equipment vibration amplitude sequence, and aligns and splices the order generation time and service priority on the time axis to obtain the normalized feature sequence. The data retrieval component extracts time-aligned operating current and equipment vibration amplitude sequences from multi-source raw data sequences. The mean normalization unit performs mean normalization on the operating current and equipment vibration amplitude sequences to obtain a dimensionless index with a single feature dimension. The mean normalization unit obtains the current value at the current moment, the mean of the operating current sequence within the time window, and the range of the operating current sequence. It calculates the difference between the current value and the mean of the operating current sequence within the time window to obtain the current centering deviation, and then calculates the ratio of the current centering deviation to the range of the operating current sequence to obtain the normalized operating current data. Simultaneously, the mean normalization unit obtains the current equipment vibration amplitude value, the mean of the equipment vibration amplitude sequence within the time window, and the range of the equipment vibration amplitude sequence. It calculates the difference between the current vibration amplitude value and the mean of the equipment vibration amplitude sequence to obtain the vibration centering deviation, and then calculates the ratio of the vibration centering deviation to the range of the equipment vibration amplitude sequence to obtain the normalized vibration amplitude data. When the operating current range or the equipment vibration amplitude range is extremely small and equal to 0, the mean normalization calculation unit directly outputs the normalization result as 0. Retrieving the aforementioned output value, the current operating current is 45 amps. The mean normalization calculation unit calculates the mean of the operating current sequence within the current time window to be 40 amps. Within the time window, the maximum current value of 50 amps and the minimum current value of 30 amps are extracted, and the difference is calculated to obtain the operating current sequence range of 20 amps. Substituting this into the aforementioned calculation, the current centering deviation is 5 amps. Dividing this deviation by the operating current sequence range of 20 amps yields the normalized operating current data as 0.25. Based on the aforementioned output, the current equipment vibration amplitude is 2.5 mm / s, the average vibration amplitude sequence within the current time window is 2.0 mm / s, and the difference between the maximum amplitude of 3.0 mm / s and the minimum amplitude of 1.0 mm / s yields a vibration amplitude sequence range of 2.0 mm / s. The calculated vibration centering deviation is 0.5 mm / s. Dividing this by the vibration amplitude sequence range of 2.0 mm / s yields a normalized vibration amplitude data of 0.25. The time-series stitching unit aligns and stitches the order generation time and service priority along the time axis. The time-series stitching unit extracts the time offset of the reference zero point as the aligned order generation time feature and combines it with the aforementioned quantized value 3, stitching it into the normalized feature sequence.

[0024] The sequence construction submodule extracts normalized operating current data and normalized vibration amplitude data based on the normalized feature sequence, and calls the order generation time and service priority identifier after time axis alignment to perform field-level splicing and sequential arrangement operations to generate a multi-source state sequence. The feature extraction and parsing unit accurately extracts the normalized operating current data value of 0.25 and the normalized vibration amplitude data value of 0.25 from the received normalized feature sequence. Simultaneously, the feature extraction and parsing unit calls the time offset value of 1005 milliseconds corresponding to the order generation time after time axis alignment, and the quantized value of 3 corresponding to the service priority identifier. The field concatenation and sorting unit performs field-level concatenation and sequential arrangement operations on the extracted discrete data points to obtain a structured multi-dimensional state row vector. The field concatenation and sorting unit embeds a fixed field aggregation template, which specifies that the data arrangement order is forcibly set as the normalized operating current data field, the normalized vibration amplitude data field, the order time offset field, and the service priority quantized value field. The field concatenation and sorting unit writes the cross-sectional data of each dimension one by one according to the above arrangement order. After extracting all the aforementioned numerical calculation results, they are sequentially filled into the corresponding position cells, generating a single-row structured data set that sequentially aggregates the values ​​0.25, 0.25, 1005, and 3. The state matrix generation unit vertically stacks single-row structured data sets from multiple consecutive sampling periods along the time dimension to obtain a multi-source state sequence. The state matrix generation unit sets a preset sequence window length, calculated based on the average time taken for a single feeding action at the workstation. Equipment monitoring logs show that the time taken for a typical single operation is consistently around 2 seconds. Based on the aforementioned 50Hz baseline sampling frequency, the state matrix generation unit keeps the sequence window length constant at 100 sampling points. The state matrix generation unit continuously acquires newly generated single-row structured datasets, merges them, and pushes them into a memory queue. When the number of sequentially stacked data sets in the queue reaches exactly 100, the state matrix generation unit triggers a data sequence encapsulation action, generating a two-dimensional multi-source state sequence matrix with 100 rows and 4 columns.

[0025] Specifically, such as Figure 2 , 4 As shown, the mutation detection module includes: The amplitude difference construction submodule obtains the current amplitude difference and vibration amplitude difference between adjacent times based on the multi-source state sequence, obtains the current amplitude difference sequence by differentiating the current sampling under the continuous time index, obtains the vibration amplitude difference sequence by differentiating the vibration sampling at the corresponding time, and aligns it with the current amplitude difference sequence in time to obtain the amplitude difference combination vector. Based on the storage bus parsing of the pre-encapsulated two-dimensional multi-source state sequence matrix, the current state data vector and the previous adjacent state data vector are extracted row by row in chronological order. The sequence difference operation unit, considering the occasional high-frequency glitches in the device sensor signals, calls a limiting filter to perform boundary checks on the extracted values. The sequence difference operation unit retrieves the current-time normalized operating current data from the current-time state data vector and simultaneously retrieves the previous-time normalized operating current data from the previous adjacent state data vector. The current-time normalized operating current data is subtracted from the previous-time normalized operating current data to calculate the single-point amplitude difference of the current. Simultaneously, the sequence difference operation unit retrieves the current-time normalized vibration amplitude data from the current-time state data vector and the previous-time normalized vibration amplitude data from the previous adjacent state data vector. The current-time normalized vibration amplitude data is subtracted from the previous-time normalized vibration amplitude data to calculate the single-point amplitude difference of the vibration. The above difference processing is performed continuously on 100 consecutive rows of data in the multi-source state sequence matrix, and then summarized to form the current amplitude difference sequence and vibration amplitude difference sequence corresponding to the time index order. Substituting the previous output values ​​for calculation, the state data reading component extracts the normalized operating current data of 0.25 and the normalized vibration amplitude data of 0.25 at the current time offset of 1005 milliseconds, and simultaneously extracts the normalized operating current data of 0.20 and the normalized vibration amplitude data of 0.15 at the previous time offset of 1000 milliseconds. The sequence difference operation unit subtracts the value of 0.20 from the value of 0.25 to obtain the single-point current amplitude difference value of 0.05, and similarly subtracts the value of 0.15 from the value of 0.25 to obtain the single-point vibration amplitude difference value of 0.10. The amplitude difference alignment and combination unit extracts the same time index label, and concatenates the calculated single-point current amplitude difference value and the single-point vibration amplitude difference value horizontally to construct an amplitude difference combination vector containing the dynamic difference dimension.

[0026] The distance measurement submodule, based on the amplitude difference combination vector, calls the difference and sum of squares operation between vector components to construct the distance calculation expression. It multiplies the current amplitude difference and vibration amplitude difference under the same time index by the corresponding weight coefficients to unify the dimensions, and then performs differential square accumulation and square root to obtain the feature distance sequence. The internal data structure of the amplitude difference combination vector is analyzed to accurately read the single-point amplitude difference of current and vibration under the same time index label. The metric expression calculation unit multiplies the extracted single-point amplitude difference of current with itself to obtain the squared value of current difference, and simultaneously multiplies the extracted single-point amplitude difference of vibration with itself to obtain the squared value of vibration difference. The metric expression calculation unit then sums the squared values ​​of current and vibration differences to obtain the cumulative sum of squared differences across dimensions. The metric expression calculation unit further calls the square root function to calculate the square root of the cumulative sum of squared differences, obtaining the feature distance value that can characterize the intensity of transient joint changes of multidimensional signals. The sequence summation component arranges and combines the continuously calculated feature distance values ​​vertically according to the ascending order of time index, finally generating a one-dimensional feature distance sequence. Verification calculation is performed using the actual parameter values ​​output above. The metric expression calculation unit retrieves the single-point amplitude difference of current (0.05) and vibration (0.10) at a time offset of 1005 milliseconds. The measurement and expression calculation unit multiplies the single-point current amplitude difference of 0.05 by itself to obtain the squared value of the current difference, 0.0025, and multiplies the single-point vibration amplitude difference of 0.10 by itself to obtain the squared value of the vibration difference, 0.0100. Then, the values ​​0.0025 and 0.0100 are added together to obtain the cumulative sum of the squared differences, 0.0125. The measurement and expression calculation unit performs a square root operation on the value 0.0125 to finally obtain the feature distance value of 0.1118.

[0027] The anomaly marking submodule obtains the distance at multiple time points based on the feature distance sequence and compares it point by point with the preset feature distance benchmark threshold. It marks the time indexes that exceed the feature distance benchmark threshold as an anomaly index set, performs time mapping on the anomaly index set, and generates a state change time series. The feature distance values ​​corresponding to each time point are extracted one by one from the feature distance sequence and input into a comparator to perform a point-by-point comparison operation with a preset feature distance benchmark threshold. When the extracted feature distance value is strictly greater than the preset feature distance benchmark threshold, the threshold comparison analysis component triggers a marking action and stores the time index to which the value belongs in the anomaly index set. The setting of the feature distance benchmark threshold relies on Gaussian distribution fitting of feature distance data from 10,000 status samples collected during 6 consecutive months of normal and fault-free operation of the workstation equipment. Experimental statistical calculations confirm that the upper limit critical value of the feature distance in the 99% confidence interval is constant at 0.1000. Therefore, the threshold comparison analysis component precisely sets the feature distance benchmark threshold to 0.1000. Substituting the results obtained from the previous deduction, the threshold comparison analysis component extracts the feature distance value of 0.1118 at the current time offset of 1005 milliseconds and performs a numerical comparison command with it and the feature distance benchmark threshold of 0.1000. Since the feature distance value of 0.1118 is greater than the feature distance baseline threshold of 0.1000, this result indicates that the current physical operating state of the equipment has undergone a sudden and drastic change that exceeds the normal fluctuation tolerance. This means that the time point is confirmed as the starting point of the sudden equipment failure. Therefore, the threshold comparison analysis component writes the time index corresponding to the time offset of 1005 milliseconds into the anomaly index set. The time mapping encapsulation unit retrieves the original time axis aligned record dictionary, reverses the relative time index number of the records in the anomaly index set to convert it into the absolute millisecond-level time parameter of the business issuance timestamp, and outputs the state change time sequence in chronological order.

[0028] Specifically, such as Figure 2 , 5 As shown, the tensor rearrangement module includes: The coordinate extraction submodule extracts the time coordinates of mutation points and non-mutation points based on the state mutation time and multi-source state sequence. It matches the state mutation time and locates the time in the multi-source state sequence. It records the positions of mutation points and non-mutation points according to the time and normalizes the coordinates to obtain a set of time coordinates. The generated state mutation time series and multi-source state sequence matrix are obtained. The time analysis and localization component extracts the abnormal absolute timestamp values ​​contained in the state mutation time series. Combined with the preceding parameters, the state mutation timestamp is confirmed to be 1005 milliseconds. The time analysis and localization component executes a traversal comparison command on the time axis of the multi-source state sequence matrix. When the timestamp recorded in the matrix is ​​strictly consistent with the state mutation timestamp, it is confirmed as the mutation point time coordinate, and the remaining timestamps in the matrix are confirmed as non-mutation point time coordinates. The time analysis and localization component records the corresponding row index position of the mutation point time coordinate in the multi-source state sequence matrix as 10, and records the row index positions of the non-mutation point time coordinates as 1 to 9 and 11 to 100 respectively. The coordinate normalization processing unit obtains the absolute time values ​​of each recorded time coordinate, calculates the absolute time difference by subtracting the set starting time coordinate value of the matrix from the currently extracted absolute time value, and then divides the absolute time difference by the set range time span value of the matrix to finally calculate the normalized time coordinate value. The sequence start time coordinate is set to 955 milliseconds, and the sequence end time coordinate is set to 1450 milliseconds. The coordinate normalization unit subtracts these two values ​​to obtain the range time span value of 495 milliseconds. Then, the abrupt change point time coordinate (1005 milliseconds) is subtracted from the start time coordinate value of 955 milliseconds to obtain the absolute time difference value of 50 milliseconds. This absolute time difference value of 50 milliseconds is divided by the range time span value of 495 milliseconds to calculate the normalized time coordinate value of 0.1010. The coordinate normalization unit then combines the normalized time coordinates of the abrupt change point with the calculated normalized time coordinates of all non-abrupt change points and encapsulates them to output a time coordinate set.

[0029] Table 2. Normalized Time Coordinate Records for Mutation Points and Non-Mutation Points 1 955 955 495 0.0000 Non-mutation point 10 1005 955 495 0.1010 Mutation point Table 2 shows the parameter correspondence mapping records before and after normalization processing for various time nodes in the coordinate extraction stage.

[0030] The priority assignment submodule assigns priority numbers to the time coordinates of mutation points based on the time coordinate set, performs numerical mapping on the location indices of multiple mutation points and assigns incremental numbers according to the order of mutation occurrence, sorts the numbering results, and generates a priority number sequence. The time-series incremental mapping component extracts all normalized time coordinate values ​​containing mutation point status markers from the time coordinate set. When multiple mutation points are triggered within a single fixed sequence window, the component obtains the magnitude of the normalized time coordinate values ​​for each mutation point and defines the mutation evolution stage based on the value range. Values ​​between 0 and 0.3 are defined as very early mutations, values ​​between 0.3 and 0.7 as mid-stage mutations, and values ​​between 0.7 and 1.0 as late-stage mutations. The system's built-in initial priority base value is always 1. Strictly following the ascending chronological order of the normalized time coordinate values ​​of each mutation point, the component assigns this base value to the earliest appearing mutation point on the timeline. Subsequently, it performs an incremental mapping calculation, incrementing by 1 for each subsequent mutation point, to obtain the corresponding priority number value. The coordinate parsing and reading unit extracts the normalized time coordinate value of the first mutation point as 0.1010. This value is determined to be in the very early mutation range, and the time-series incremental mapping component assigns it a priority number of 1. Next, the normalized time coordinate value of the second mutation point within the same window is extracted as 0.4500. This value is determined to be in the middle mutation range, and the time-series incremental mapping component increments the previous base value by 1, resulting in a priority number of 2 for this mutation point. The sequence sorting and encapsulation unit obtains each generated priority number value, binds it to the corresponding original time coordinate row index, and reconstructs the data storage structure according to the ascending order of priority numbers to obtain the priority number sequence.

[0031] The index rearrangement submodule rearranges the time indexes of mutation points and non-mutation points based on the priority number sequence. It inserts the mutation point indexes into the beginning of the time series in numerical order, while padding and arranging the non-mutation point indexes in their original order. It then performs index splicing and tensor structure reorganization to obtain rearranged tensor data. The internal structure of the priority number sequence is analyzed, and the original time coordinate row indices of all mutation points bound to priority number values ​​are extracted. Simultaneously, the original time coordinate row indices of non-mutation points without bound numbers are also extracted. The time series reconstruction and splicing component constructs a blank one-dimensional array vector based on the set length of the original multi-source state sequence. Then, according to the magnitude of the priority number values ​​bound to the mutation points, and in a strictly increasing order starting from 1, the original time coordinate row indices of each mutation point are sequentially filled into the absolute starting position of the blank one-dimensional array vector. After all mutation point row indices have been written, the time series reconstruction and splicing component obtains the original time coordinate row indices of non-mutation points and, strictly following their original time-increasing order, fills the remaining physical spaces in the blank one-dimensional array vector to obtain the rearranged index vector. Based on the actual parameter values ​​output above, the row indices for mutation points are confirmed to be 10 and 45, while the row indices for non-mutation points cover 1 to 9 and other natural values. The temporal reconstruction splicing component places the mutation row index values ​​10 and 45 into the first and second positions of the rearranged index vector, respectively. Subsequently, starting from the third position, the non-mutation row index value 1 is sequentially filled in, the fourth position is filled in with the value 2, and so on to complete the splicing and filling. The tensor reconstruction output unit calls the original multi-source state sequence matrix stored locally, extracts the row number arrangement order recorded in the rearranged index vector, extracts the corresponding data feature vectors from the multi-source state sequence matrix row by row according to the rearranged order, and performs a vertical stacking splicing operation in the newly allocated memory space to obtain the rearranged tensor data.

[0032] Specifically, such as Figure 2 , 6 As shown, the collaborative management and control module includes: The logistics overlap submodule, based on rearranged tensor data, obtains business entity identifiers and logistics paths, detects the transmission records between multiple path points and accumulates them to obtain the total logistics transmission volume, calculates the time intersection length by combining the task time series and multiplies it by the duration conversion factor related to the total logistics transmission volume, and generates the entity overlap duration set. The business entity identifier code and corresponding logistics path node list are extracted from the constructed rearranged tensor data. The flow frequency statistics component scans the logistics path node list, extracts the single transfer record value between any two adjacent path points, and continuously accumulates all the extracted single transfer record values ​​to calculate the total logistics transfer volume. For example, if the extracted single transfer record value from logistics node 1 to logistics node 2 is 5 times, and the transfer record value from node 2 to node 3 is 8 times, the flow frequency statistics component adds the value 5 and the value 8 to calculate that the total logistics transfer volume corresponding to this business entity identifier is 13 times. The time intersection calculation unit extracts the start and end values ​​of the task time series of the current business entity, and also extracts the start and end values ​​of the task time series of the comparison entity, obtaining the start absolute timestamp and end absolute timestamp of the overlap between the two. The time intersection calculation unit subtracts the start absolute timestamp from the end absolute timestamp to calculate the time intersection length. If the overlap start absolute timestamp is read as 1200 seconds and the end absolute timestamp is read as 1500 seconds, the difference is calculated to obtain a time intersection length of 300 seconds. The fusion processing component combines and binds the total output of logistics delivery with the time intersection length to form a two-dimensional set of physical overlap durations and stores it in the storage bus.

[0033] The connectivity clustering submodule, based on the entity overlap duration set, calls the product operation of the total logistics transfer volume and the intersection length to obtain the connectivity, sorts them in descending order according to the connectivity and divides them into multiple data clusters in equal quantities, extracts the entity spatial coordinate sequence within the cluster and performs centering adjustment to obtain the entity coordinate sequence of the same cluster. The system reads the total logistics transfer volume and time intersection length under the bound state from the entity overlap duration set. The connectivity calculation unit directly multiplies the total logistics transfer volume by the time intersection length to calculate the connectivity value reflecting the degree of collaboration between entities. In the calculation, the total logistics transfer volume (13 times) is multiplied by the time intersection length (300 seconds) to obtain a connectivity value of 3900. The descending clustering component extracts all calculated connectivity values, executes a value comparison instruction, and rearranges all connectivity values ​​into a sequence according to the descending order rule. Subsequently, the descending clustering component reads the total length of the sequence, divides it by the preset fixed number of data clusters to obtain the single cluster data capacity, and performs an equal-step truncation operation along the sequence from front to back according to the single cluster data capacity, dividing it into multiple data clusters. The total length of the connectivity value sequence is set to 1000 entity objects, and the preset fixed number of data clusters is 10. The descending clustering component divides the data to obtain a single cluster data capacity of 100, and then assigns the first 100 entities to the first data cluster. Since the connectivity value of 3900 calculated in the previous step is ranked 45th in the sequence, this result indicates that the interaction frequency between the current entities is at the forefront of the active range. For the step result, this means that the entity has been precisely assigned to the first data cluster. The coordinate centering adjustment unit extracts the absolute spatial x-coordinate and absolute spatial y-coordinate of all business entities within the first data cluster, and calculates the arithmetic mean of the x-coordinates and y-coordinates respectively to obtain the cluster center coordinates. The coordinate centering adjustment unit subtracts the arithmetic mean of the x-coordinates of each entity from the absolute spatial x-coordinate to obtain the centered x-coordinate, and similarly obtains the centered y-coordinate. These centered coordinate sets are then packaged and output as a sequence of entity coordinates within the same cluster.

[0034] The instruction generation submodule obtains the device slice status and calculates the remaining duration, calls the coordinate sequence of entities in the same cluster to match the associated slice number, performs difference calculation on the remaining duration of slices in the same cluster to extract the remaining time difference sequence, and reassembles the instruction fields according to the slice number and the time difference sequence to generate a collaborative control instruction set. The system reads the running configuration log of the currently running device and extracts the upper limit of the slice lifecycle and the current running time from the device slice status parameters. The status parsing and calculation unit subtracts the current running time from the upper limit of the slice lifecycle to calculate the remaining duration of the device slice. In the actual running environment, the extracted upper limit of the slice lifecycle is 3600 seconds, and the current running time is 2100 seconds. The status parsing and calculation unit subtracts these two to obtain a remaining duration of 1500 seconds. The number association component reads the coordinate sequence of entities in the same cluster. Based on the centralized horizontal and vertical coordinate features recorded in the sequence, it calls the underlying slice registry dictionary and performs string comparison matching one by one to extract the unique associated slice number bound to the entity object in the same cluster. The duration difference extraction component selects the remaining duration data output by all devices in the same cluster, extracts the largest remaining duration value as the baseline duration, subtracts the remaining duration of each device in the same cluster from the baseline duration, calculates the remaining time difference value for each device, and arranges all remaining time difference values ​​according to the associated slice number to form a remaining time difference sequence. If the maximum base duration within the same cluster is confirmed to be 2000 seconds, and the remaining duration of a certain device is 1500 seconds, the duration difference extraction component calculates the remaining time difference value of the device to be 500 seconds. The instruction reassembly and splicing unit extracts the associated slice number and the corresponding remaining time difference value, writes the associated slice number into the positioning field of the protocol header of the control instruction packet, and writes the remaining time difference value into the payload value field of the control instruction packet, completing the concatenation and splicing of the underlying binary fields to generate a collaborative management and control instruction set.

[0035] Specifically, such as Figure 2 , 7 As shown, the reallocation module includes: The parameter parsing submodule, based on the collaborative management and control instruction set, obtains accelerated production compensation parameters and logistics capacity scheduling parameters, aligns fields according to time tags and equipment identifiers, and performs index mapping and element concatenation on accelerated production compensation and logistics capacity scheduling based on equipment numbers to establish a joint parameter mapping set. The system reads the generated collaborative control instruction set, parses the binary header sequence of the instruction payload data packet, and extracts the accelerated production compensation parameter values ​​and logistics capacity scheduling parameter values ​​for multiple device objects. The accelerated production compensation parameter value is quantified by subtracting the factory rated speed from the current actual operating speed obtained by the device's built-in speed sensor at a sampling frequency of 50 Hz. The logistics capacity scheduling parameter value is quantified by multiplying the number of currently available transport vehicles retrieved from the scheduling interface by the standard load constant per vehicle. For example, if the operating speed of device number 1001 is extracted to be 1200 rpm and the rated speed is 1000 rpm, the instruction unpacking unit calculates the difference, resulting in a speed difference of 200, which is taken as its accelerated production compensation parameter value. Simultaneously, the number of available transport vehicles for this node is extracted to be 5, and the standard load constant per vehicle is 10 tons; multiplying these together, the logistics capacity scheduling parameter value is 50 tons. Similarly, the speed difference extracted from equipment number 1002 yields an accelerated production compensation parameter value of 150, corresponding to a logistics capacity scheduling parameter value of 40 tons. The accelerated production compensation parameter value extracted from equipment number 1003 is 250, corresponding to a logistics capacity scheduling parameter value of 60 tons. The field timestamp alignment component scans the absolute values ​​of the timestamps inherent in the extracted parameters and the equipment identifier codes existing as hexadecimal strings. Parameters with identical timestamps under the same equipment identifier code are extracted into the same temporary cache queue. The index mapping splicing unit reads the corresponding equipment number and performs a sequential write operation on the data in the temporary cache queue associated with the same equipment number, arranging them in the physical contiguous locations in memory. This horizontal splicing and fusion of the two types of parameters is completed in the underlying data matrix, establishing and outputting a structured parameter joint mapping set.

[0036] The threshold verification submodule calls the parameter joint mapping set, calculates the difference between each accelerated production compensation parameter and the preset equipment rated load benchmark threshold, determines the sign attribute of the difference and marks the parameter items that exceed the equipment rated load benchmark threshold, records the corresponding parameter index and the over-limit boundary, and obtains the load over-limit boundary set. The underlying storage parameter union mapping set is invoked, and the accelerated production compensation parameter values ​​attached to each device number are extracted sequentially. The preset rated load baseline threshold for each device is strictly set to a fixed safe value of 100, based on the maximum safe speed redundancy specified on the device's nameplate and the mechanical wear attenuation coefficient after 1000 hours of continuous trouble-free operation. The load difference calculation component extracts the accelerated production compensation parameter values ​​item by item, and directly subtracts the preset rated load baseline threshold from each accelerated production compensation parameter value to calculate the load difference. In the actual calculation, multiple sets of device data obtained from the aforementioned parameter parsing are substituted. For device number 1001, the load difference calculation component subtracts the baseline threshold 100 from the parameter value 200 to calculate a load difference of 100; for device number 1002, the parameter value 150 subtracts the baseline threshold 100 to calculate a load difference of 50; and for device number 1003, the parameter value 250 subtracts the baseline threshold 100 to calculate a load difference of 150. The boundary determination and recording unit extracts the aforementioned load difference and executes a binary sign bit detection instruction. If the load difference is greater than zero (i.e., the sign attribute is determined to be positive), it is determined that the accelerated production compensation parameter value has exceeded the upper limit range of safe equipment operation, and the parameter item is marked as a high-risk over-limit state. The load differences of the above three devices are all greater than zero, indicating that multiple devices are currently in a high-risk overload operation range. This means that the parameters of these devices are forcibly entered into the abnormal monitoring queue. The boundary determination and recording unit extracts the physical storage row number index position of the over-limit state parameter item in the parameter joint mapping set matrix and defines the calculated corresponding load difference as the over-limit boundary value. Subsequently, the boundary determination and recording unit performs key-value pair mapping and binding operations between the physical storage row number index position and each over-limit boundary value, packaging them to obtain a load over-limit boundary set.

[0037] The capacity reconfiguration submodule extracts the overload parameters and corresponding boundaries from the load overload boundary set, performs boundary trimming on the overload parameters, splices and reassembles the trimmed accelerated production compensation parameters and logistics capacity scheduling parameters, performs capacity allocation ratio multiplication calculation, and generates a collaborative management and control redistribution instruction set. The system reads the physical storage row number index and the over-limit boundary value corresponding to each over-limit state from the load over-limit boundary set, and backtracks to the parameter joint mapping set based on this index position to accurately extract the original over-limit accelerated production compensation parameter value. The boundary trimming component subtracts the over-limit boundary value from the over-limit accelerated production compensation parameter value to calculate the trimmed accelerated production compensation parameter value. This forced deduction operation directly reverts the equipment's operating parameters to the safe rated range. Combining the aforementioned calculation results, for equipment number 1001, the over-limit accelerated production compensation parameter value 200 is subtracted from the over-limit boundary value 100 to obtain a trimmed accelerated production compensation parameter value of 100. Similarly, for equipment 1002, 150 is subtracted from 50 to obtain a value of 100, and for equipment 1003, 250 is subtracted from 150 to obtain a value of 100. The capacity allocation and reorganization unit extracts each trimmed accelerated production compensation parameter value and calls the total accelerated production compensation parameter value obtained by summing in the global operation log and sets it to 1000. The capacity allocation and reorganization unit divides the reduced accelerated production compensation parameter values ​​for each piece of equipment by the total accelerated production compensation parameter value of 1000 to calculate the capacity allocation ratio representing the relative weight. The capacity allocation ratio for the three pieces of equipment is calculated to be 0.10. The unit extracts the original logistics capacity scheduling parameter values ​​and multiplies them by the corresponding capacity allocation ratio values ​​to calculate the reconstructed capacity allocation values. For equipment 1001, multiplying 50 tons by 0.10 yields a reconstructed capacity allocation value of 5.0 tons; for equipment 1002, multiplying 40 tons by 0.10 yields 4.0 tons; and for equipment 1003, multiplying 60 tons by 0.10 yields 6.0 tons. The unit then concatenates these reconstructed capacity allocation values ​​with the corresponding equipment numbers using the underlying data frames to generate a collaborative control and reallocation instruction set.

[0038] Table 3 Detailed Record of Capacity Reconfiguration Allocation Parameters 1001 200 100 100 0.10 50 tons 5.0 tons 1002 150 50 100 0.10 40 tons 4.0 tons 1003 250 150 100 0.10 60 tons 6.0 tons 1004 180 80 100 0.10 45 tons 4.5 tons 1005 220 120 100 0.10 55 tons 5.5 tons As shown in Table 3, the evolution of compensation parameters for multiple over-limit devices before and after the forced boundary pruning process is presented in detail, as well as the numerical results of the reconfigured capacity allocation derived from the final mapping.

[0039] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A prototype system for collaborative management and control of manufacturing and services based on supply chain integration, characterized in that: The system includes: The data analysis module acquires operating current, equipment vibration amplitude, order generation time, and service priority through sensors at the production workstations in the supply chain and order terminals. It normalizes the operating current and equipment vibration amplitude, aligns the order generation time and service priority time, generates a multi-source state sequence, and transmits it to the mutation detection module. The mutation detection module obtains the difference in current amplitude and vibration amplitude between adjacent times in the multi-source state sequence and calculates the feature distance. It then marks abnormal time points by combining the preset feature distance benchmark threshold, generates state mutation time, and transmits it to the tensor rearrangement module. The tensor rearrangement module extracts the mutation point and non-mutation point times based on the state mutation time, assigns priority values ​​to the mutation point times and rearranges the mutation point and non-mutation point times, generates rearranged tensor data and transmits it to the collaborative management module. The collaborative management module, based on the rearranged tensor data, obtains the total logistics delivery volume, task overlap duration, and equipment slice status. It performs business connectivity clustering on the total logistics delivery volume and task overlap duration, maps the spatial coordinates of business entities in the same cluster, calculates the remaining time difference in combination with the equipment slice status, and generates a collaborative management instruction set.

2. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 1, characterized in that, The multi-source state sequence includes current normalization, vibration normalization, and time alignment stamp; the state mutation time includes anomaly time index, mutation amplitude marker, and time location label; the rearranged tensor data includes mutation priority weight, non-mutation order index, and time rearrangement sequence; and the collaborative management and control instruction set includes service connectivity cluster identifier, spatial mapping coordinates, remaining time difference, and device slice code.

3. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 1, characterized in that, The data analysis module includes: The signal analysis submodule acquires sensor signals from production workstations in the supply chain and order terminal records, collects operating current sequences and equipment vibration amplitude sequences, monitors order generation timestamps and service priority identifier sequences, and performs time stamp matching on multiple types of data according to a unified time scale to obtain multi-source raw data sequences. The normalization processing submodule calls the running current sequence and the equipment vibration amplitude sequence based on the multi-source original data sequence, performs mean normalization calculation on the running current sequence and the equipment vibration amplitude sequence, and aligns and splices the order generation time and service priority on the time axis to obtain the normalized feature sequence. The sequence construction submodule extracts normalized operating current data and normalized vibration amplitude data based on the normalized feature sequence, and calls the order generation time and service priority identifier after time axis alignment to perform field-level splicing and sequential arrangement operations to generate a multi-source state sequence.

4. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 1, characterized in that, The mutation detection module includes: The amplitude difference construction submodule obtains the current amplitude difference and vibration amplitude difference between adjacent times based on the multi-source state sequence, obtains the current amplitude difference sequence by differentiating the current sampling under the continuous time index, obtains the vibration amplitude difference sequence by differentiating the vibration sampling at the corresponding time, and aligns it with the current amplitude difference sequence in time to obtain the amplitude difference combination vector. The distance measurement submodule, based on the amplitude difference combination vector, calls the difference and sum of squares operation between vector components to construct a distance calculation expression. It multiplies the current amplitude difference and vibration amplitude difference under the same time index by the corresponding weight coefficients to unify the dimensions, and then performs differential square accumulation and square root to obtain the feature distance sequence. The anomaly marking submodule obtains the distance at multiple time points based on the feature distance sequence and compares it point by point with a preset feature distance benchmark threshold. Time indices that exceed the feature distance benchmark threshold are marked as anomaly index sets. The anomaly index sets are then time-mapped to generate a state change time series.

5. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 4, characterized in that, The characteristic distance reference threshold is obtained by acquiring the operating current and vibration amplitude of the equipment within the calibration period, extracting the difference in current amplitude and the difference in vibration amplitude between adjacent times, multiplying them by the corresponding weight coefficients, accumulating the difference squares and taking the square root to generate a calibration distance sequence, calculating the arithmetic mean and standard deviation of the calibration distance sequence, and summing the arithmetic mean and the standard deviation by adding a preset multiple to determine the distance.

6. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 1, characterized in that, The tensor rearrangement module includes: The coordinate extraction submodule extracts the time coordinates of the mutation point and the non-mutation point based on the state mutation time and the multi-source state sequence, matches the state mutation time and locates the time in the multi-source state sequence, records the positions of the mutation point and the non-mutation point according to the time and normalizes the coordinates to obtain a set of time coordinates. The priority assignment submodule assigns priority numbers to the time coordinates of mutation points according to the time coordinate set, performs numerical mapping on the location indices of multiple mutation points and assigns incremental numbers according to the order of mutation occurrence, sorts the numbering results, and generates a priority number sequence. The index rearrangement submodule rearranges the time coordinates of mutation points and non-mutation points based on the priority number sequence. It inserts the mutation point indexes into the beginning of the time series in numerical order, while arranging the non-mutation point indexes in their original order. It then performs index splicing and tensor structure recombination to obtain rearranged tensor data.

7. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 1, characterized in that, The collaborative management module includes: The logistics overlap submodule, based on the rearranged tensor data, obtains the business entity identifier and logistics path, detects the transmission records between multiple path points and accumulates them to obtain the total logistics transmission volume, calculates the time intersection length by combining the task time series and multiplies it by the duration conversion factor related to the total logistics transmission volume, and generates the entity overlap duration set. The connectivity clustering submodule, based on the entity overlap duration set, calls the product operation of the total logistics transfer volume and the intersection length to obtain the connectivity, sorts them in descending order according to the connectivity and divides them into multiple data clusters in equal quantities, extracts the entity spatial coordinate sequence within the cluster and performs centering adjustment to obtain the entity coordinate sequence of the same cluster. The instruction generation submodule obtains the device slice status and calculates the remaining duration, calls the coordinate sequence of the same cluster entity to match the associated slice number, performs difference calculation on the remaining duration of the slices in the same cluster to extract the remaining time difference sequence, and reassembles the instruction fields according to the slice number and the time difference sequence to generate a collaborative control instruction set.

8. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 1, characterized in that, The system also includes: The reallocation module obtains accelerated production compensation parameters and logistics capacity scheduling parameters based on the collaborative management and control instruction set, compares them with preset equipment load capacity thresholds, truncates the accelerated production compensation parameters, and simultaneously reallocates capacity from the collaborative management and control instruction set to generate a collaborative management and control reallocation instruction set. The collaborative management and reallocation instruction set includes compensation truncation parameters, capacity allocation ratios, and scheduling priority codes.

9. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 8, characterized in that, The reallocation module includes: The parameter parsing submodule, based on the collaborative management and control instruction set, obtains accelerated production compensation parameters and logistics capacity scheduling parameters, aligns fields according to time tags and equipment identifiers, and performs index mapping and element concatenation of accelerated production compensation and logistics capacity scheduling according to equipment number to establish a joint parameter mapping set; The threshold verification submodule calls the parameter joint mapping set, maps the accelerated production compensation parameters to equivalent load values ​​through preset conversion rules, calculates the difference with the preset equipment rated load benchmark threshold, determines the sign attribute of the difference and marks the parameter items that exceed the equipment rated load benchmark threshold, and obtains the load over-limit boundary set. The capacity reconfiguration submodule extracts the over-limit parameters and corresponding boundaries from the load over-limit boundary set, performs boundary trimming on the over-limit parameters, splices and reassembles the trimmed accelerated production compensation parameters and logistics capacity scheduling parameters, and performs capacity allocation ratio multiplication calculation to generate a collaborative management and control redistribution instruction set.

10. The prototype system for collaborative management and control of manufacturing and services based on supply chain integration as described in claim 9, characterized in that, The rated load reference threshold of the equipment is obtained by acquiring the operating power and rated current within the equipment calibration period, and using the operating power divided by the rated voltage to calculate the rated current, etc., in accordance with the laws of physics, to form a calibration load sequence. The arithmetic mean and standard deviation of the calibration load sequence are calculated, the standard deviation is multiplied by a preset multiple to obtain the deviation, and the arithmetic mean and the deviation are summed to determine the value.

Images