A numerical control machine tool production efficiency statistical system

Abstract

The application relates to the technical field of intelligent manufacturing, in particular to a numerical control machine tool production efficiency statistical system, which comprises a data acquisition and correlation unit, which is used for acquiring and correlating multi-dimensional state data from a numerical control machine tool; a switching feature vectorization unit, which is used for receiving the multi-dimensional state data acquired by the data acquisition and correlation unit; a transient parameter prediction unit, which is used for predicting an initial efficiency drop amplitude and a feature recovery time based on a switching feature vector constructed by the switching feature vectorization unit; a transient loss quantification unit, which is used for calculating a transient efficiency loss according to the initial efficiency drop amplitude and the feature recovery time predicted by the transient parameter prediction unit; and a scheduling sequence optimization unit, which is used for taking the transient efficiency loss calculated by the transient loss quantification unit as a decision basis to solve an optimal task execution sequence. The application can improve the actual output efficiency of the numerical control machine tool and the overall equipment utilization rate in the dynamic production scene of small batches and multiple varieties.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, specifically to a CNC machine tool production efficiency statistics system. Background Technology

[0002] In the current trend of small-batch, multi-variety manufacturing, CNC machine tools need to frequently switch machining tasks. Traditional efficiency statistics methods, such as calculating the overall efficiency of the equipment or the total machining time, mainly focus on the machine tool's performance in a steady state, often neglecting the crucial implicit cost of switching tasks. When a machine tool switches from task A to task B, the system experiences a transient instability phase involving tool changing, thermal balance adjustment, and program loading. During this period, the machine tool's actual efficiency drops significantly before gradually recovering to the steady-state level of the new task. The efficiency loss caused by this recovery process accumulates in environments with frequent switching, severely impacting actual output. Existing scheduling algorithms often treat switching time as a fixed constant, failing to accurately reflect the differentiated costs between different task switching pairs (such as A->B and A->C), resulting in scheduling schemes that are not actually optimal. Therefore, designing a system that can accurately quantify and predict this transient efficiency loss through statistical analysis of historical data, and using this result to guide production scheduling, is a pressing technical problem in this field. Summary of the Invention

[0003] The purpose of this invention is to provide a CNC machine tool production efficiency statistics system, which solves the problems existing in the background art.

[0004] To solve the above-mentioned technical problems, the present invention provides a CNC machine tool production efficiency statistics system, comprising:

[0005] The data acquisition and association unit is used to acquire and associate multi-dimensional status data from CNC machine tools;

[0006] The switching feature vectorization unit is used to receive multi-dimensional state data collected by the data acquisition and association unit and construct a switching feature vector.

[0007] The transient parameter prediction unit is used to predict the initial efficiency drop magnitude and feature recovery time based on the switching feature vector constructed by the switching feature vectorization unit.

[0008] The transient loss quantization unit is used to calculate the transient efficiency loss based on the initial efficiency drop amplitude and feature recovery time predicted by the transient parameter prediction unit.

[0009] The scheduling sequence optimization unit is used to solve for the optimal task execution sequence by taking the transient efficiency loss calculated by the transient loss quantization unit as the basis for decision-making.

[0010] Preferably, the multidimensional status data includes machine tool status data, tool information, and program information.

[0011] Preferably, the switching feature vectorization unit is used to obtain the spindle temperature, tool type code, and cutting load at the end of the source task and the start of the target task to construct a state feature vector.

[0012] Preferably, the switching feature vectorization unit is further used to obtain whether the fixture needs to be changed and the target task's thermal stability time requirement, in order to construct a switching process complexity feature vector;

[0013] The switching feature vector is constructed by combining the state feature vector built by the switching feature vectorization unit and the switching process complexity feature vector.

[0014] Preferably, the transient parameter prediction unit calculates the initial efficiency drop magnitude based on the state feature vector constructed by the switching feature vectorization unit through a linear regression model.

[0015] Preferably, the transient parameter prediction unit calculates the feature recovery time using an exponential model based on the switching process complexity feature vector constructed by the switching feature vectorization unit.

[0016] Preferably, the transient loss quantization unit multiplies the initial efficiency drop amplitude predicted by the transient parameter prediction unit with the feature recovery time, and sets the resulting product as the transient efficiency loss;

[0017] Transient efficiency loss characterizes the equivalent production downtime caused by task switching.

[0018] Preferably, the solution process for the scheduling sequence optimization unit is as follows:

[0019] The transient efficiency loss calculated by the transient loss quantization unit during switching between any two tasks is used to form a cost matrix.

[0020] Set the arrangement of the set of tasks to be scheduled as a task scheduling sequence;

[0021] The transient efficiency loss of switching between all adjacent tasks in the task scheduling sequence is summed to obtain the total cumulative time loss.

[0022] A heuristic optimization algorithm is used to solve the cost matrix with the goal of minimizing the cumulative total time loss, thereby obtaining the optimal task execution sequence.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] (1) Accurate quantification and prediction of transient efficiency loss during task switching is achieved. By setting up a data acquisition association unit, a switching feature vectorization unit, a transient parameter prediction unit and a transient loss quantification unit, a complete data-driven prediction link is constructed. This link can predict the initial efficiency drop amplitude and feature recovery time based on the multi-dimensional state data collected and associated from the machine tool by constructing a switching feature vector and using linear regression and exponential models.

[0025] (2) This invention proposes and verifies a derivable transient efficiency loss calculation model, which transforms the complex dynamic process into a single indicator with clear physical meaning that can be used for decision-making. It converts the total efficiency loss caused by a task switch into a specific production downtime and provides managers and scheduling systems with an intuitive and comparable cost metric.

[0026] (3) By using a heuristic optimization algorithm to solve this matrix, a scheduling scheme that may not be optimal locally, but can avoid several severe efficiency drops in the entire task sequence, thereby minimizing cumulative losses, can be generated. Ultimately, this achieves the beneficial effect of maximizing the actual output of equipment in a small-batch, multi-variety production mode. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0028] Figure 1 This is a logic block diagram of the system of the present invention;

[0029] Figure 2 This is a logical block diagram of the multidimensional state data of the present invention;

[0030] Figure 3 This is a logic block diagram of the feature vectorization unit switching in this invention;

[0031] Figure 4 This is a logic block diagram for the construction of the feature vector switching in this invention. Detailed Implementation

[0032] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0033] Please see Figure 1 This invention provides a CNC machine tool production efficiency statistics system, comprising:

[0034] The data acquisition and association unit is used to acquire and associate multi-dimensional status data from CNC machine tools;

[0035] The switching feature vectorization unit is used to receive multi-dimensional state data collected by the data acquisition and association unit and construct a switching feature vector.

[0036] The transient parameter prediction unit is used to predict the initial efficiency drop magnitude and feature recovery time based on the switching feature vector constructed by the switching feature vectorization unit.

[0037] The transient loss quantization unit is used to calculate the transient efficiency loss based on the initial efficiency drop amplitude and feature recovery time predicted by the transient parameter prediction unit.

[0038] The scheduling sequence optimization unit is used to solve for the optimal task execution sequence by taking the transient efficiency loss calculated by the transient loss quantification unit as the decision basis.

[0039] This embodiment discloses a CNC machine tool production efficiency statistical system. Its core objective is to accurately quantify and optimize efficiency losses caused by task switching based on prediction. The system's internal operating logic begins with the real-time collection and synchronization of multi-dimensional machine tool status data by a data acquisition and association unit. These raw data streams are then processed by a switching feature vectorization unit, transforming them into structured mathematical inputs suitable for model analysis. Based on this structured input, a transient parameter prediction unit uses a statistical learning model to accurately predict two core parameters defining the dynamic process of efficiency recovery: the initial efficiency drop amplitude and the feature recovery time. Subsequently, a transient loss quantification unit mathematically integrates these two predicted parameters to calculate the specific loss amount caused by each task switch. This loss value is given a clear physical meaning. The term "switching cost" refers to an equivalent period of production downtime. As the final decision-making stage, the scheduling sequence optimization unit uses these quantified loss values ​​as cost inputs to the optimization algorithm. By solving a combinatorial optimization problem aimed at minimizing cumulative switching losses, it outputs an optimal task execution sequence. This entire process constructs a complete technical closed loop from data perception to optimization decision-making. It fundamentally abandons the rough assumption in traditional scheduling methods that treats switching costs as a fixed constant. By deeply mining historical data and establishing predictive models, production scheduling decisions can accurately adapt to the objectively existing differences in costs between different task switching pairs. The ultimate technical effect is to significantly improve the actual output efficiency of CNC machine tools and the overall equipment utilization rate in dynamic production scenarios with small batches and multiple varieties.

[0040] Example 2

[0041] Please see Figure 2 Multidimensional status data includes machine tool status data, tool information, and program information;

[0042] The switching feature vectorization unit is used to obtain the spindle temperature, tool type code, and cutting load at the end of the source task and the start of the target task in order to construct the state feature vector;

[0043] In a specific application scenario of this embodiment, the data acquisition and association unit serves as the data foundation of the entire system. Its key role is to continuously collect and form a time-synchronized multidimensional data stream from the machine tool CNC system, external sensors, and the manufacturing execution system. To achieve time synchronization, this system sets up a data aggregation node, using the internal clock of the machine tool CNC system as the reference time source. All data packets sent by external sensors and the manufacturing execution system (MES) contain their respective generation timestamps. After receiving the data, the data aggregation node performs periodic time synchronization with each data source using Network Time Protocol (NTP) or Precision Time Protocol (PTP), calculating the data transmission delay between each data source and the reference time source. During data association, the timestamps of non-reference sources are corrected according to the following formula: ,in This is the corrected timestamp. The original timestamp of the data source. This is used to calculate the average network transmission latency from the data source to the aggregation node. This method ensures precise alignment of all state data on the timeline, providing a data foundation for accurately capturing the state at the moment of task switching.

[0044] These multidimensional state data form the foundation for all subsequent quantitative analyses, and their specific content includes machine tool state data, such as the actual spindle speed. With spindle load Tool information, such as the unique identifier of the current tool. With tool wear monitoring value And program information, such as the G-code segment being executed. G-code is an international standard term in the field of CNC machine tools, specifically referring to the instruction language that controls the machining process, such as the movement, speed, and cutting tools of the machine tool. Based on this data, the function of the switching feature vectorization unit is specified: it accurately captures the time node of each task switching event and extracts the source task respectively. End moment and target task Key physical quantities at the start, such as the real-time temperature of the spindle, the type code of the tool currently being used, and the average load during the cutting process, are organized and encoded into a state feature vector. The state feature vector It should include at least the following dimensions:

[0045] Spindle temperature: Floating-point value used directly;

[0046] Tool type coding: One-hot encoding is used. If the system has... If the tool type is specified, then the code is a... A dimension vector, with the current tool type having a dimension of 1 and the rest having 0;

[0047] Cutting load: defined as the average spindle load percentage within 2 seconds before the source task ends and within 2 seconds after the target task begins;

[0048] Speed ​​deviation: defined as the root mean square value of the difference between the actual spindle speed and the program command speed (S command) within a specific time window before and after task switching;

[0049] Through the above-mentioned clear feature definitions and quantification methods, the consistency and repeatability of the state feature vector construction are guaranteed;

[0050] By transforming discrete, multi-source, heterogeneous time-series data into structured state feature vectors, this unit provides quantifiable and comparable inputs for subsequent prediction models, enabling the system to identify and measure the magnitude of state transitions before and after different task switching, which is a necessary prerequisite for accurately predicting switching costs.

[0051] Example 3

[0052] Please see Figure 3 The switching feature vectorization unit is further used to obtain whether the fixture needs to be changed and the target task's thermal stability time requirement, in order to construct a feature vector of the switching process complexity.

[0053] Furthermore, by combining the state feature vector constructed from the switching feature vectorization unit and the switching process complexity feature vector, a switching feature vector is constructed.

[0054] To achieve a more refined and comprehensive prediction of handover costs, the handover feature vectorization unit in this embodiment is further equipped with a handover process complexity feature vector, in addition to constructing a state feature vector. The unit's functional design not only focuses on the physical state differences before and after the switchover, but also delves into the operational complexity of the switchover process itself. It obtains Boolean variables such as whether the switchover involves changing fixtures (e.g., a value of 1 indicates a change is needed, and a value of 0 indicates no change is required) and the specific requirements of the target task for machine tool thermal balance, such as a specific thermal stabilization time requirement. This non-physical state information is integrated into the switchover process complexity feature vector. Ultimately, this unit will characterize the state feature vector representing the changes in physical state. The feature vector representing the complexity of the switching process. By combining these features, a more comprehensive feature vector can be constructed that fully describes the characteristics of the switching event. The effectiveness of this technique is that it enables subsequent prediction models to not only perceive changes in physical state but also understand the inherent complexity of the operation process. This allows for more accurate prediction of nonlinear time delays caused by complex operations, greatly improving the fidelity and generalization ability of the entire system's prediction model.

[0055] Example 4

[0056] Please see Figure 4 The transient parameter prediction unit calculates the initial efficiency drop magnitude based on the state feature vector constructed by the switching feature vectorization unit through a linear regression model.

[0057] The transient parameter prediction unit calculates the feature recovery time using an exponential model based on the switching process complexity feature vector constructed by the switching feature vectorization unit.

[0058] In the core prediction stage of this embodiment, the transient parameter prediction unit is designed to integrate two independent but complementary regression sub-models, which are used to predict the initial efficiency drop magnitude. and feature recovery time ;

[0059] The first sub-model is a linear regression model that receives a state feature vector constructed from the switching feature vectorization unit and calculates the initial efficiency drop magnitude according to the following formula:

[0060] ;

[0061] The underlying logic is that the degree of efficiency impact at the moment of task switching is strongly correlated with the magnitude of the difference in the physical state of the machine tool before and after the switching; a linear regression model is used to establish this proportional relationship in a direct and physically interpretable way.

[0062] In this formula, Indicates from the source task Switch to target task The predicted initial efficiency drop magnitude is a dimensionless number.

[0063] and These represent the state feature vectors at the end of the source task and the beginning of the target task, respectively.

[0064] It is the L1 norm of the difference between two vectors, used to quantify the total magnitude of the state change;

[0065] It is the feature weight vector. These are bias terms, and both are model parameters obtained through supervised learning training on a large amount of historical switching event data;

[0066] To obtain training labels for supervised learning, historical data needs to be processed to pinpoint the actual initial efficiency drop magnitude that occurred during each switching event. and feature recovery time The calibration process is as follows:

[0067] Step 1: Define a proxy metric for instantaneous efficiency. The ratio of spindle load to the steady-state load of the task will be used as a proxy metric for instantaneous efficiency. ,Right now ,in Let be the real-time spindle load at time t. The steady-state spindle load is statistically obtained from historical data for this task.

[0068] Step 2: Determine the historical steady-state efficiency and initial efficiency for a complete historical task. By analyzing its load curve, and using methods such as the sliding window standard deviation method, when the standard deviation of the load is lower than a preset threshold within a sufficiently long time window (e.g., 10 seconds), the machine tool is considered to have entered a steady state. The average load within this window is the mean load. After the handover occurs ( The efficiency proxy index of the first sampling point is the initial efficiency. .

[0069] Step 3: Calculate the historical initial efficiency drop magnitude. The actual efficiency drop magnitude label for this switchover is defined as: .

[0070] Step 4: Calculate historical feature recovery time. This involves using a series of instantaneous efficiency proxy metrics collected after the handover. As a time series, it is fitted nonlinearly, with the objective function being a first-order exponential recovery model: The parameters obtained by solving using the least squares method This is the feature recovery time stamp for this switch. .

[0071] By following the steps above, a set of deterministic rules can be generated for each history switching event. Label pairs are used to construct a complete, labeled dataset for model training.

[0072] To ensure dimensional consistency, The dimensions of each element are set to be the reciprocals of their corresponding characteristic dimensions, such that... The product is dimensionless, and the result is the same as the product on the left. Dimensional matching; This model enables the transient parameter prediction unit to accurately predict the instantaneous decrease in efficiency when switching occurs, based on the specific physical state parameters of the two tasks.

[0073] The second sub-model is an exponential model that receives a feature vector of the switching process complexity and calculates the feature recovery time according to the following formula:

[0074] ;

[0075] The motivation for designing this model is that recovery time, as a physical quantity that is always positive, has complex nonlinear coupling relationships among its influencing factors; using an exponential model can naturally guarantee the non-negativity of the predicted output and effectively capture this complex composite effect.

[0076] A unit time reference or base time constant;

[0077] In this formula, This represents the predicted feature recovery time, and its physical dimension is time.

[0078] It is a feature vector describing the operational complexity of the switching process;

[0079] and These are model parameters, the values ​​of which are determined through supervised learning training on the recovery times actually observed in historical data.

[0080] Represents the natural exponential function;

[0081] To ensure dimensional consistency, the exponent term It must be dimensionless; this is achieved by making the weights dimensionless during training. Dimensions and corresponding characteristics The dimensions of the terms are reciprocals of each other, and the bias term is made to... To achieve a dimensionless output, the dimensionless output of the exponential function is multiplied by a unit time base normalized in the model, thus assigning the final output... With the correct time dimension, the collaborative work of these two sub-models enables the system to fully characterize the entire transient process from two orthogonal dimensions: how deep the fall and how fast the recovery, providing all the necessary dynamic parameters for the final accurate quantification of the total switching cost.

[0082] Example 5

[0083] The transient loss quantization unit multiplies the initial efficiency drop amplitude predicted by the transient parameter prediction unit with the characteristic recovery time, and sets the resulting product as the transient efficiency loss; the transient efficiency loss characterizes the equivalent production downtime caused by task switching;

[0084] In this embodiment, the transient loss quantization unit performs a crucial transformation step: converting a dynamic, time-varying efficiency recovery process into a static, single cost. This transformation is based on a mathematical model describing the efficiency recovery curve, which first defines the transient efficiency at any given time.

[0085] ;

[0086] The model was established based on the discovery, after statistical analysis of a large amount of historical production data, that the efficiency recovery curve after task switching can generally be fitted with a first-order exponential recovery function with high precision.

[0087] In this formula, It is the time after the handover occurs. Predicted transient efficiency over time;

[0088] The target task The theoretical steady-state efficiency, which is obtained through statistical historical data, serves as a priori knowledge input; The method for determining it is as follows:

[0089] For historical tasks: iterate through all tasks in the database. Based on the historical machining records, the steady-state spindle load for each machining record was calculated using the aforementioned method of 'calibrating historical steady-state efficiency'. All historical records The values ​​are statistically averaged and then discarded. After removing outliers (outside three standard deviations), the final average is taken as the theoretical steady-state load of the task. In this system, the steady-state efficiency is... It is normalized to 1, while transient efficiency is manifested as a deviation from this '1'.

[0090] For a completely new task: If a task If no historical data is available, prediction is performed using the K-Nearest Neighbor (k-NN) algorithm based on task process parameters (such as material, tool, depth of cut, and width of cut). The algorithm searches the historical task database for data related to the task. The process parameters are most similar For each task, a weighted average of its theoretical steady-state load values ​​is calculated (the weights can be proportional to the similarity), and this result is used as the basis for a new task. The theoretical steady-state load is estimated, thereby completing the construction of the efficiency model.

[0091] and These are the outputs of the aforementioned transient parameter prediction unit;

[0092] Based on this efficiency curve model, the transient loss quantization unit calculates the difference between the ideal steady-state efficiency and the actual transient efficiency. arrive Integrate to calculate the total efficiency loss. The calculation formula and derivation results are as follows:

[0093] ;

[0094] The physical meaning of this formula derives from the definition of total production loss, which is the area enclosed by the actual efficiency curve and the ideal steady-state efficiency level; its final analytical solution... This reveals a profound intrinsic relationship; the transient loss quantization unit performs this calculation, and its core technological effect is to accurately quantize a complex dynamic process into a single numerical value with clear physical meaning. The physical dimension of this value is time, which intuitively represents the equivalent production downtime caused by this task switch. This quantification method provides a clear, summable, and comparable cost metric for subsequent scheduling optimization, and is the technical foundation for realizing optimization decisions based on quantitative analysis.

[0095] Example 6

[0096] The solution process for the scheduling sequence optimization unit is as follows:

[0097] The transient efficiency loss calculated by the transient loss quantization unit during switching between any two tasks is used to form a cost matrix.

[0098] Set the arrangement of the set of tasks to be scheduled as a task scheduling sequence;

[0099] The transient efficiency loss of switching between all adjacent tasks in the task scheduling sequence is summed to obtain the total cumulative time loss.

[0100] A heuristic optimization algorithm is used to solve the cost matrix with the objective of minimizing the cumulative total time loss, thereby obtaining the optimal task execution sequence;

[0101] In the final decision-making phase of the system, the scheduling sequence optimization unit takes over the transient efficiency loss among all pairs of tasks to be scheduled, calculated by the transient loss quantification unit. Based on this, a system optimization process is executed; the first step of this process is to schedule the sequence optimization unit to switch all tasks (e.g., from task A to B, A to C, B to A, etc.) and their corresponding quantization losses. The values ​​are organized into an asymmetric cost matrix. ,in This cost matrix constitutes the mathematical model of the entire optimization problem; subsequently, this unit will set the tasks to be scheduled. An arrangement Define a specific task scheduling sequence and establish an objective function to evaluate the total cost of any given sequence P:

[0102] ;

[0103] The construction of this objective function is based on the modeling idea of ​​the Traveling Salesman Problem (TSP) in the field of combinatorial optimization. Its purpose is to find an execution path that traverses all task nodes and minimizes the total switching cost.

[0104] In this formula, It is the cumulative total time loss of the scheduling sequence P, and its physical dimension is time;

[0105] From the sequence Switch to the first task The equivalent time loss for each task;

[0106] Since solving this asymmetric traveling salesman problem precisely is an NP-hard problem, the computational cost is too high when the number of tasks is large. Therefore, the scheduling sequence optimization unit in this embodiment embeds an efficient heuristic optimization algorithm, such as the ant colony algorithm or simulated annealing algorithm; this unit aims to minimize the cumulative total time loss. To optimize the objective, the cost matrix is ​​iteratively solved within an acceptable computational time, ultimately outputting a near-optimal task execution sequence. The ultimate technical effect of this optimization process is that the system can provide the production workshop with a production dispatch sequence that is data-driven and model-predicted and verified, which can minimize the hidden time costs caused by frequent switching, thereby directly translating into higher overall equipment efficiency and stronger production rhythm adaptability, providing solid and reliable technical support for lean production in the intelligent manufacturing environment.

[0107] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art can use the above-disclosed technical content to make changes or modifications to equivalent embodiments and apply them to other fields. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention shall still fall within the protection scope of the technical solution of the present invention.

Claims

1. A CNC machine tool production efficiency statistics system, characterized in that, include: The data acquisition and association unit is used to acquire and associate multi-dimensional status data from CNC machine tools; The switching feature vectorization unit is used to receive multi-dimensional state data collected by the data acquisition and association unit and construct a switching feature vector. The transient parameter prediction unit is used to predict the initial efficiency drop magnitude and feature recovery time based on the switching feature vector constructed by the switching feature vectorization unit. The transient loss quantization unit is used to calculate the transient efficiency loss based on the initial efficiency drop amplitude and feature recovery time predicted by the transient parameter prediction unit. The scheduling sequence optimization unit is used to solve for the optimal task execution sequence by taking the transient efficiency loss calculated by the transient loss quantification unit as the decision basis. The solution process for the scheduling sequence optimization unit is as follows: The transient efficiency loss calculated by the transient loss quantization unit during switching between any two tasks is used to form a cost matrix. Set the arrangement of the set of tasks to be scheduled as a task scheduling sequence; The transient efficiency loss of switching between all adjacent tasks in the task scheduling sequence is summed to obtain the total cumulative time loss. A heuristic optimization algorithm is used to solve the cost matrix with the goal of minimizing the cumulative total time loss, thereby obtaining the optimal task execution sequence.

2. The CNC machine tool production efficiency statistics system according to claim 1, characterized in that, Multidimensional status data includes machine tool status data, tool information, and program information.

3. The CNC machine tool production efficiency statistics system according to claim 1, characterized in that, The switching feature vectorization unit is used to obtain the spindle temperature, tool type code, and cutting load at the end of the source task and the start of the target task to construct the state feature vector.

4. The CNC machine tool production efficiency statistics system according to claim 1, characterized in that, The switching feature vectorization unit is used to obtain whether to change the fixture and the target task's thermal stability time requirement, in order to construct a feature vector of the switching process complexity; The switching feature vector is constructed by combining the state feature vector built by the switching feature vectorization unit and the switching process complexity feature vector.

5. The CNC machine tool production efficiency statistics system according to claim 1, characterized in that, The transient parameter prediction unit calculates the initial efficiency drop magnitude using a linear regression model based on the state feature vector constructed by the switching feature vectorization unit.

6. The CNC machine tool production efficiency statistics system according to claim 1, characterized in that, The transient parameter prediction unit calculates the feature recovery time using an exponential model based on the feature vector of switching process complexity constructed by the switching feature vectorization unit.

7. The CNC machine tool production efficiency statistics system according to claim 1, characterized in that, The transient loss quantization unit multiplies the initial efficiency drop magnitude predicted by the transient parameter prediction unit with the feature recovery time, and sets the resulting product as the transient efficiency loss. Transient efficiency loss characterizes the equivalent production downtime caused by task switching.

Images