An experimental process high time resolution recording and automatic experimental report generation method and system

Abstract

The application provides a kind of experimental process high time resolution record and automatic experiment report generation method and system, it is related to experimental data management technical field, the method includes: with second level, sub-second level or higher frequency continuous acquisition in experimental process multi-source state data, multi-source state data includes the image or video data of experimental object, experimental environment or device running data, experimental execution action data and experimental decision and path selection data, and add time stamp for each multi-source state data;The multi-source state data collected is associated with corresponding experimental behavior, experimental intervention action and experimental decision, to obtain the mapping relationship between the experimental state and behavior, to obtain the associated record data;The associated record data is organized in time sequence, to build the multidimensional time series experimental data set containing state change, behavior change and decision change.The application improves the completeness, traceability, reproducibility and analysis depth of experimental results.

Description

Technical Field

[0001] This invention relates to the field of experimental data management and automated experimental information processing technology, and in particular to a method and system for high-temporal-resolution recording, correlation analysis, and automatic generation of experimental reports. Background Technology

[0002] When conducting liquid-phase synthesis experiments of novel nanocatalytic materials in university chemical engineering laboratories, researchers used a traditional recording method combining manual timed recording with a simple device log. Every 5 minutes, they manually recorded the operating parameters of the device, such as the reactor temperature and stirring rate. Actions such as feeding and pH adjustment were scattered in the experimental logbook. The phased detection results of the liquid chromatograph were also stored separately on the device terminal. This not only failed to capture the key sub-second transient temperature rise during the experiment, but also failed to establish a traceable correlation between pH adjustment actions and changes in the material's crystal form. After the experiment, several hours were spent manually integrating the various data scattered in the logbook and on the device terminal to write the experimental report. Ultimately, repeated experiments failed to reproduce the same catalytic material performance indicators due to the lack of core data. The above application scenario is only an example and does not constitute a limitation on the types and scenarios of experiments applicable to this invention.

[0003] This experimental case highlights the technical shortcomings of traditional experimental recording methods. Their time resolution is only at the minute level, making it impossible to capture transient changes and key moments in the experiment. At the same time, the recorded content is fragmented, with experimental status, operational behavior, and test results stored in a scattered manner without a unified structure. There is a lack of traceable correlation between experimental behavior and experimental results, and the generation of experimental reports relies entirely on manual processing, which is inefficient and prone to data omissions due to manual operation. Ultimately, this seriously affects the reproducibility of experimental results and the depth of analysis. Summary of the Invention

[0004] This invention provides a method and system for high temporal resolution recording of experimental processes and automatic generation of experimental reports, which improves the reproducibility and analytical depth of experimental results. It enables high temporal resolution continuous recording of experimental processes, unified correlation of multi-source data, automatic identification of key nodes and abnormal events, and automatic generation of structured experimental reports, thereby improving the reproducibility, traceability, and analytical depth of experimental results.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: In a first aspect, a method for high-temporal-resolution recording of experimental processes and automatic generation of experimental reports, the method comprising: Multi-source state data during the experiment is continuously collected at a frequency of seconds, sub-seconds or higher. The multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision and path selection data. A timestamp is added to each piece of multi-source state data. The collected multi-source state data is associated with the corresponding experimental behaviors, experimental intervention actions and experimental decisions to construct a mapping relationship between experimental states and behaviors, so as to obtain associated record data. Organize the associated record data in chronological order to construct a multidimensional time series experimental dataset that includes state changes, behavior changes, and decision changes; Analyze multidimensional time series experimental datasets to identify key nodes, key changes, or anomalous events in the experimental process, and extract features of the experimental results; Based on the characteristics of the experimental results, a structured description of the experimental results and an experimental report are automatically generated. The experimental report includes at least the experimental conditions, experimental process, key nodes, result analysis and conclusions, and establishes a correlation and reference relationship with the experimental process data, key node data and / or abnormal event data.

[0006] Furthermore, multi-source state data is continuously acquired during the experiment at a frequency of seconds, sub-seconds, or higher. This multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision-making and path selection data. A timestamp is added to each piece of multi-source state data, including: Images, videos, or microscopic imaging data of the experimental object are continuously acquired at a frequency of seconds, sub-seconds, or higher using an image sensor, and a timestamp is added to each data point; different types of data, such as experimental environment or device operation, can be acquired using the same or different sampling frequencies, aligned using a unified time reference, and a time stamp is established for each data point; experimental execution action data are synchronously recorded through experimental execution control, and a timestamp is added to each data point; Key result information is obtained from external experimental equipment or modules. The acquisition methods include display content recognition, interface reading, log parsing, data synchronization, or a combination thereof, and a timestamp is added to each piece of information. Experimental decision and path selection data are recorded synchronously through experimental decision-making, and a timestamp is added to each piece of data to obtain multi-source state data with timestamps.

[0007] Furthermore, the collected multi-source state data is correlated with corresponding experimental behaviors, intervention actions, and decisions to construct a mapping relationship between experimental states and behaviors, thereby obtaining correlated record data, including: A unique experimental task identifier is generated for each experimental process; all collected multi-source state data, as well as experimental behaviors, intervention actions, and decision-making data during the experimental process, are associated with the experimental task identifier. Record the control behaviors, experimental intervention actions or parameter adjustment information, parameter values ​​before and after adjustment and the basis for adjustment related to changes in experimental state during the experiment, and associate the parameter adjustment instructions, parameter values ​​before and after adjustment and the basis for adjustment with the corresponding state data; Record the state data, execution parameters, and response actions before and after the occurrence of the abnormal event, and associate the state data, execution parameters, and response actions before and after the occurrence of the abnormal event with the experimental task identifier to obtain associated record data containing the mapping relationship between state and behavior.

[0008] Furthermore, the associated recorded data are organized chronologically to construct a multidimensional time-series experimental dataset containing state changes, behavioral changes, and decision changes, including: The associated record data is arranged in chronological order according to timestamps to form a continuous and traceable experimental process data chain; Based on the experimental process data chain, multiple time series are constructed, corresponding to the change trajectory of experimental state parameters, the execution sequence of experimental behavior, and the triggering node of experimental decision, respectively, to obtain a multi-dimensional time series experimental dataset.

[0009] Furthermore, based on the experimental process data chain, multiple time series are constructed, corresponding to the change trajectory of experimental state parameters, the execution sequence of experimental behaviors, and the triggering nodes of experimental decisions, respectively, to obtain a multi-dimensional time series experimental dataset, including: Extract all data related to experimental state parameters from the experimental process data chain and arrange them in timestamp order to obtain the time series of experimental state parameter change trajectories; extract all data related to experimental behavior execution from the experimental process data chain and arrange them in timestamp order to form the time series of experimental behavior execution timing; extract all data related to experimental decision triggering from the experimental process data chain and arrange them in timestamp order to form the time series of experimental decision triggering nodes. The time series of experimental state parameter change trajectories, experimental behavior execution time sequences, and experimental decision triggering nodes are integrated and aligned according to a unified time axis to generate a multi-dimensional time series experimental dataset.

[0010] Furthermore, the multidimensional time series experimental dataset is analyzed to identify key nodes, key changes, or anomalous events in the experimental process, and to extract features of the experimental results, including: Extract time series data from a multidimensional time series experimental dataset, and analyze the change trajectory of experimental state parameters, the execution sequence of experimental behaviors, and the temporal correlation characteristics between the triggering nodes of experimental decisions; Based on the aforementioned temporal correlation features, the moments when experimental state parameters change or deviate from the expected trend are identified and marked as key nodes; based on the temporal correlation features, the intervals of abnormal fluctuations in experimental state caused by experimental behavior or decision triggers are identified and marked as abnormal events. Based on the marked key nodes and abnormal events, the correlation, influence, or temporal response relationships between state changes, behavior execution, and decision triggering during the experiment are analyzed to determine the final state, stage output results, or comprehensive results of the experiment, which are then used as the experimental results. Feature parameters are extracted from the experimental results. These feature parameters include at least the state values ​​at key nodes, the rate of change, the duration of abnormal events, and the scope of their impact, in order to obtain the characteristics of the experimental results.

[0011] Furthermore, based on the characteristics of the experimental results, a structured description of the experimental results and an experimental report are automatically generated. The experimental report includes at least the experimental conditions, experimental process, key milestones, result analysis, and conclusions, and establishes a correlation and reference relationship with the experimental process data, key milestone data, and / or abnormal event data, including: Based on the characteristics of the experimental results, a framework for the experimental report is constructed, which includes a summary of experimental conditions, an overview of the experimental process, a list of key nodes, a results analysis section, and a conclusion section. Extract experimental starting conditions and environmental parameters from the multidimensional time series experimental dataset and populate the experimental condition summary; extract the main behavioral sequences and state change trajectories during the experimental execution process from the multidimensional time series experimental dataset to generate an overview of the experimental process; Organize key nodes and abnormal events into a list of key nodes in chronological order, and associate each node with a corresponding state data fragment and behavior description; Based on the characteristics of the experimental results, the correlation, influence, or time-series response relationship between state changes and behavioral decisions during the experiment is analyzed, and the results analysis text is generated. The experimental conclusions are generated by integrating the experimental process and results analysis, and all parts are integrated into a structured experimental report.

[0012] Secondly, a system for high-temporal-resolution recording of experimental processes and automatic generation of experimental reports includes: The acquisition module is used to continuously collect multi-source state data during the experiment at a frequency of seconds, sub-seconds, or higher. The multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision and path selection data. A timestamp is added to each piece of multi-source state data. The association module is used to associate the collected multi-source state data with the corresponding experimental behaviors, experimental intervention actions and experimental decisions, and to build a mapping relationship between experimental states and behaviors in order to obtain associated record data. The building module is used to organize the associated record data in chronological order and build a multidimensional time series experimental dataset that includes state changes, behavior changes, and decision changes; The identification module is used to analyze multidimensional time series experimental datasets, identify key nodes, key changes or abnormal events in the experimental process, and extract experimental result features. The processing module is used to automatically generate a structured description of experimental results and an experimental report based on the characteristics of the experimental results. The experimental report includes at least the experimental conditions, experimental process, key nodes, result analysis and conclusions, and establishes a correlation and reference relationship with the experimental process data, key node data and / or abnormal event data.

[0013] Thirdly, a computing device, comprising: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0014] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0015] The above-described solution of the present invention has at least the following beneficial effects: This approach effectively overcomes the technical challenges of traditional experimental recording methods. It employs techniques such as continuously acquiring time-stamped multi-source experimental state data at second / sub-second or higher frequencies; assigning unique task identifiers to experiments and associating state data with experimental behaviors / interventions / decisions; constructing multi-dimensional time-series experimental datasets containing state / behavior / decision changes along a unified timeline; non-manually identifying key nodes and anomalies based on temporal correlation features and analyzing correlations, influence relationships, or temporal response relationships; and automatically generating structured experimental reports with data fragment associations based on experimental result characteristics. These techniques address the problems of traditional methods, including low temporal resolution failing to capture transient changes, fragmented and unstructured recording content, lack of traceability between experimental behaviors and results, and reliance on manual, inefficient, and subjective report generation. Ultimately, this approach achieves high-granularity, continuous, and traceable structured recording throughout the experimental process, establishes a clear causal mapping between experimental state changes and behavioral decisions, accurately identifies key experimental nodes and abnormal events, improves the reproducibility and analytical depth of experimental results, reduces the cost of manual recording and report writing, and provides high-quality data support for experimental knowledge accumulation, pattern discovery, and model training. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for high-time-resolution recording of experimental processes and automatic generation of experimental reports, provided by an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of an experimental process high time resolution recording and automatic experimental report generation system provided by an embodiment of the present invention. Detailed Implementation

[0018] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0019] like Figure 1 As shown, an embodiment of the present invention proposes a method for high temporal resolution recording of experimental processes and automatic generation of experimental reports, the method comprising the following steps: Step 1: Continuously collect multi-source state data during the experiment at a frequency of seconds, sub-seconds or higher. The multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision and path selection data. Add a timestamp to each piece of multi-source state data. Step 2: Associate the collected multi-source state data with the corresponding experimental behaviors, experimental intervention actions and experimental decisions to construct a mapping relationship between experimental states and behaviors in order to obtain associated record data; Step 3: Organize the associated record data in chronological order to construct a multidimensional time series experimental dataset that includes state changes, behavior changes, and decision changes; Step 4: Analyze the multidimensional time series experimental dataset, identify key nodes, key changes or abnormal events in the experimental process, and extract the features of the experimental results; Step 5: Based on the characteristics of the experimental results, automatically generate a structured description of the experimental results and an experimental report. The experimental report shall include at least the experimental conditions, experimental process, key nodes, result analysis and conclusions, and establish a correlation and reference relationship with the experimental process data, key node data and / or abnormal event data.

[0020] In an embodiment of the present invention, In a preferred embodiment of the present invention, step 1 above may include: Step 1.1: Continuously acquire images, videos, or microscopic imaging data of the experimental object at a frequency of seconds, sub-seconds, or higher using an image sensor, and add a timestamp to each data point. Different types of data, such as experimental environment or device operation data, can be acquired using the same or different sampling frequencies, aligned with a unified time reference, and a time stamp is established for each data point. Synchronously record experimental execution action data through experimental execution control, and add a timestamp to each data point. Specifically, before acquiring high-temporal-resolution data during the experiment, complete the parameter configuration and linkage debugging of the acquisition equipment. First, determine a unified high-frequency acquisition standard, and set the acquisition interval to seconds or sub-seconds according to the experimental type and data capture requirements. The sampling frequency is calculated using the formula: Sampling Frequency = 1 / Sampling Interval. If the sub-second sampling interval is set to 0.5 seconds, the sampling frequency is 1 ÷ 0.5 = 2 Hz. If the second-level sampling interval is set to 1 second, the sampling frequency is 1 ÷ 1 = 1 Hz. The sampling frequencies of the image sensor, environmental sensor, external device information acquisition module, and experimental data recording system are all adjusted to this standard to ensure time synchronization of data acquisition across all dimensions. Simultaneously, the timestamp generation rules for all data recording terminals are configured, uniformly adopting the system's real-time time format of year-month-day hour:minute:second, millisecond, ensuring consistent timestamp format for each data entry. The above sampling frequency and timestamp generation rules are for illustrative purposes only and do not limit the scope of protection. In practical applications, the sampling interval or sampling frequency can be configured according to the experimental type and data capture requirements. The timestamp can be an absolute timestamp, a relative timestamp, a time sequence number, or a combination thereof.

[0021] Image sensors are deployed at key observation locations on the experimental object. If microscopic observation is required, a microscopic imaging component is used. After the image sensors are activated, they continuously collect images, videos, or microscopic imaging data of the experimental object at a preset acquisition frequency. For each data acquisition, a timestamp conforming to preset rules is generated and the timestamp is bound and stored with the acquired image, video, or microscopic imaging data. Simultaneously, environmental sensors are deployed at key monitoring points in the experimental environment and at the core operating parts of the experimental device. The environmental sensors continuously collect environmental data such as temperature, humidity, air pressure, and ventilation volume, as well as operating data such as rotation speed, temperature, pressure, and flow rate of the experimental device, at the same preset acquisition frequency. For each environmental or device operating data acquisition, the system synchronously generates and binds a corresponding timestamp. The experimental execution control system establishes real-time linkage with the experimental execution device. When the experimental execution device performs actions such as feeding, parameter adjustment, equipment start-up and shutdown, and process switching, the experimental execution control system immediately captures the action information, records the action type, execution parameters, and action trigger time, etc. After recording each execution action data at a preset acquisition frequency, the system generates and binds a timestamp in a unified format.

[0022] Step 1.2: Obtain key result information from external experimental equipment or modules. The acquisition methods include display content recognition, interface reading, log parsing, data synchronization, or a combination thereof, and add a timestamp to each piece of information. Experimental decision-making and path selection data are recorded synchronously through experimental decision-making, and a timestamp is added to each piece of data to obtain multi-source state data with timestamps. Specifically, this includes: aligning the image acquisition component with the display interface of the external experimental equipment and continuously acquiring real-time images of the display interface at a preset uniform acquisition frequency. The acquired display interface images are then transmitted in real-time to the optical character recognition processing module. This module first performs preprocessing on the images, including grayscale conversion, noise reduction, and character segmentation, and then performs character recognition and information extraction on the preprocessed images. It accurately extracts key result information such as detection values, experimental results, and parameter indicators from the display interface. Each extracted... The system binds key result information to the timestamp of the corresponding display interface image to ensure the time consistency of external device result information with other collected data. The experimental decision system establishes real-time data communication with the experimental decision-making end. When decision instructions, decision adjustments, experimental path selection, path switching and other decision and path selection data are generated during the experiment, the decision and path selection data shall include at least one or more of the following: decision type, decision content, trigger basis, and execution instruction. For each record of decision and path selection data, the system generates and binds a corresponding timestamp according to a unified rule. All experimental object perception data, experimental environment and device operation data, experimental execution action data, key result information of external experimental equipment, and experimental decision and path selection data that have completed timestamp binding are integrated to form a complete set of multi-source state data with timestamps.

[0023] In this embodiment of the invention, by standardizing the configuration of a unified high-frequency acquisition frequency and timestamp format, the synchronization and high temporal resolution acquisition of multi-source experimental data are achieved, overcoming the problems of low temporal resolution and asynchronous acquisition of different types of data in traditional experimental records, and ensuring the integrity of the temporal dimension of the data. The automatic acquisition and recording of core experimental data through various sensors and control systems replaces manual timed recording, reducing omissions and human errors caused by manual recording and improving the accuracy and real-time performance of data acquisition. The automated extraction and recording of results from external experimental equipment using optical character recognition technology solves the problems of manual transcription and low integration efficiency of external equipment data in traditional experiments, achieving automated acquisition of multi-source data throughout the entire experimental process. This provides a complete data foundation with a unified time stamp for the subsequent correlation of experimental data and behaviors, and the construction of time-series datasets, solving the problem of fragmented content in traditional experimental records from the source of data acquisition, and ensuring the continuity and traceability of the experimental data chain.

[0024] In a preferred embodiment of the present invention, step 2 above may include: Step 2.1: Generate a unique experimental task identifier for each experimental process. All collected multi-source state data, as well as experimental behaviors, intervention actions, and decision-making data during the experimental process, are associated with the experimental task identifier. Specifically, this includes generating a unique experimental task identifier for each experimental process. The identifier generation rule is that it can be generated based on experimental identifier information, time information, equipment information, or a combination thereof. These three elements are sequentially concatenated to form a unique identifier without repetition, ensuring that data from different experimental processes can be accurately distinguished. Subsequently, the system establishes a mapping library linking experimental task identifiers with various types of data. All collected multi-source state data with timestamps, as well as experimental behaviors, intervention actions, and decision-making data generated in real time during the experiment, are bound to the experimental task identifier in the mapping library. This achieves a unified association between all experimental-related basic data and experimental task identifiers, ensuring the uniqueness of all data attribution.

[0025] Step 2.2 records the control behaviors, experimental intervention actions, or parameter adjustment information related to changes in the experimental state during the experiment, the parameter values ​​before and after adjustment, and the basis for adjustment triggering. It also associates the parameter adjustment instructions, the parameter values ​​before and after adjustment, and the basis for adjustment triggering with the corresponding state data. Specifically, during the experiment, if the experimental execution control system issues a parameter adjustment instruction, it accurately captures and completely records the content of the parameter adjustment instruction, including the type of parameter being adjusted, the direction of adjustment, and the execution method. At the same time, it records the original value of the parameter before adjustment, the target value after adjustment, and the basis for parameter adjustment. The triggering basis includes relevant information such as the corresponding experimental state data characteristics, experimental decision instruction requirements, and trend changes in the experimental process. It also generates a unique adjustment record number for each parameter adjustment information. Next, the system matches multi-source state data within the corresponding time range based on the generation timestamp of the parameter adjustment command. The time matching range is calculated by adding the time length offset forward from the parameter adjustment command timestamp to the time length offset backward. The offset time length is determined based on the sampling frequency of the experimental data. If the sampling frequency is 1 Hz (1 data point per second) and the offset time length is set to 2 seconds, the matching range is 2 seconds before and after the command timestamp, and the corresponding number of matched state data is the sampling frequency multiplied by the total offset time, i.e., 1 × (2 + 2) = 4 data points. If the sampling frequency is 2 Hz (1 data point per 0.5 seconds) and the offset time length is 2 seconds, the number of matched state data is 2 × (2 + 2) = 8 data points. This method accurately matches the experimental state data corresponding to the parameter adjustment action, binding the parameter adjustment command, the parameter values ​​before and after adjustment, the adjustment trigger basis, and the matched corresponding state data, thus completing the accurate association mapping between parameter adjustment information and experimental state data. In practical applications, the associated time window can be flexibly configured according to the sampling strategy, data latency characteristics, or experimental scenario.

[0026] Step 2.3: Record the state data, execution parameters, and response actions before and after the occurrence of the abnormal event, and associate these data with the experimental task identifier to obtain associated record data containing the mapping relationship between state and behavior. Specifically, this includes: automatically determining the time coverage range of the abnormal event, which is the time stamp of the abnormal event detection triggered forward by a preset warning time and backward to the time stamp of the completion of the response action. The warning time is uniformly set to 5 seconds according to the experimental type. If the experimental data acquisition frequency is 1 Hz, the amount of state data that can be extracted during the abnormal warning stage is the acquisition frequency multiplied by the warning time, i.e., 1 × 5 = 5 records. Extract all experimental state data and experimental execution parameters within this time coverage range, and simultaneously record the complete response action in real time. Information such as the trigger command, execution type, execution time, and execution result of the handling action is collected, and a unique abnormal record number is generated for the abnormal event. All record information of the state data, execution parameters, and response actions before and after the abnormal event is bound in the association mapping library of the experimental task identifier, ensuring that all abnormal data belongs to the unique identifier of the corresponding experimental process. The system integrates the association data of basic data and experimental task identifier, the association data of step parameter adjustment information and corresponding state data, and the association data of abnormal data and experimental task identifier to form a complete association record data containing the mapping relationship between experimental state and experimental behavior. In practical applications, the above-mentioned warning time setting can be determined according to the experimental type, risk level, or preset rules to determine the time coverage of the abnormal event.

[0027] In this embodiment of the invention, by generating a unique experimental task identifier and achieving unified association of all experimental data, the problem of scattered storage and lack of unified attribution identifiers in traditional experimental records is overcome, thus realizing centralized management of experimental data. By fully recording complete information on parameter adjustments and accurately matching corresponding state data according to the collection frequency, a direct association between parameter adjustment behavior and experimental state changes is established, solving the problem of disconnect between experimental operations and state data in traditional records and realizing traceability of operational behavior and state changes. By defining the time coverage of abnormal events and fully recording and associating related data, the defects of incomplete records and lack of clear association of abnormal event-related data in traditional experimental records are compensated for, realizing full-process data traceability of abnormal events from early warning to handling. The finally formed associated record data establishes a clear mapping relationship between experimental state and experimental behavior, providing a traceable associated data foundation for the construction of multi-dimensional time series experimental datasets and the analysis of experimental process association, influence relationship or time-series response relationship, thus completely solving the core problem of fragmented content in traditional experimental records from the data association level.

[0028] In a preferred embodiment of the present invention, step 3 above may include: Step 3.1 involves arranging the associated record data according to the chronological order of timestamps to form a continuous and traceable experimental process data chain. Specifically, this includes: preprocessing the timestamps of all generated associated record data containing the mapping relationship between experimental states and behaviors; verifying the consistency and completeness of the timestamp format for each associated record data; removing invalid data with missing or incorrect timestamps; and retaining only valid associated record data with a unified year-month-day hour:minute:second, millisecond format. Then, the core timestamp field of each valid associated record data is extracted and used as the sole sorting criterion. All valid associated record data are arranged sequentially from the first timestamp of the experiment's start to the last timestamp of the experiment's end, using an ascending timestamp sorting method. For multiple associated record data of different types with the same timestamp, they are integrated and arranged according to a preset order of experimental state data, experimental behavior data, and experimental decision data to avoid data confusion at the same time point. After sorting, all the ordered and valid associated records are chained together. Each data record is assigned a unique data node identifier. Each data node contains the complete data content, a unified format timestamp, the associated experimental task identifier, and the association index of upstream and downstream data nodes, forming a continuous and traceable experimental process data chain. The total number of valid data nodes in this data chain is exactly equal to the total number of valid associated records after filtering. If M valid associated records are obtained after filtering, then the number of valid data nodes in the experimental process data chain is M.

[0029] Step 3.2: Based on the experimental process data chain, construct multiple time series, corresponding to the change trajectory of experimental state parameters, the execution sequence of experimental behaviors, and the triggering nodes of experimental decisions, to obtain a multi-dimensional time series experimental dataset. Specifically, this includes: constructing and integrating multiple types of time series in stages based on the data chain to obtain the final multi-dimensional time series experimental dataset; performing data classification and extraction; the first step involves traversing all valid data nodes in the experimental process data chain, accurately filtering out all experimental state parameter-related data such as the state parameters of the experimental object's image / video analysis, the experimental environment's temperature, humidity, and air pressure parameters, and the experimental device's rotation speed, pressure, and flow rate based on data attributes; reorganizing the filtered state parameter data in ascending order of timestamps to form the original data set of experimental state parameter change trajectories; if the experiment monitors K types of core state parameters, then K independent state parameter sets are formed accordingly. The first step involves traversing the data chain to obtain the original data set of state parameters. Each set contains the number of valid records for that parameter type. The second step involves iterating through all valid data nodes in the data chain, filtering out all experimental behavior-related data such as experimental execution actions, parameter adjustment actions, and response actions. Core information such as behavior type, execution time, execution parameters, and action results are extracted and sorted in ascending order by timestamp to form the original data set of experimental behavior execution. The number of data entries in this set equals the total number of experimental behaviors that occurred during the experiment. The third step involves traversing all valid data nodes in the data chain again to obtain all experimental decision-related data such as experimental decision instructions, path selection, and decision adjustments. Core information such as decision type, trigger time, decision content, and execution basis are extracted and sorted in ascending order by timestamp to form the original data set of experimental decision triggers. The number of data entries in this set equals the total number of experimental decisions generated during the experiment.

[0030] Based on three types of original data sets, independent time series were constructed. Using a unified-format timestamp as the horizontal axis and the specific values ​​of the experimental state parameters as the vertical axis, an independent time series of experimental state parameter change trajectories was constructed for each type of core state parameter, with each type of parameter corresponding to a sub-sequence. Using the unified-format timestamp as the behavior occurrence node and the complete information of the experimental behavior as the node content, an execution time series of experimental behaviors was constructed, with each node in the series corresponding to one experimental behavior. Using the unified-format timestamp as the decision trigger node and the complete information of the experimental decision as the node content, a time series of experimental decision trigger nodes was constructed, with each node in the series corresponding to one experimental decision. The multiple time series were then integrated and aligned to determine a unified time axis. The time range is from the timestamp of the first valid data node in the experimental data chain to the timestamp of the last valid data node. The total duration of this time axis is calculated as the number of milliseconds from the end timestamp to the number of milliseconds from the start timestamp. Then, the scale interval of the time axis is determined according to the high-frequency acquisition standard preset in the experiment. The scale interval is calculated by dividing 1 by the acquisition frequency in seconds. To convert it to milliseconds, the scale interval is multiplied by 1000. If the acquisition frequency of the experiment is 2 Hz, that is, 2 data points are acquired per second, then the scale interval is 1 ÷ 2 = 0.5 seconds, corresponding to 500 milliseconds. If the total duration of the time axis is 20000 milliseconds, then the total number of scales on this unified time axis is the total duration divided by the scale interval, that is, 20000 ÷ 500 = 40.

[0031] The experimental state parameter change trajectory time series, experimental behavior execution time series, and experimental decision trigger node time series constructed above are all aligned according to the scale of this unified time axis. For positions on the time axis scale that have no corresponding data, they are marked as null values ​​and the scale position is retained to ensure that the time base of the three types of time series is completely consistent. After the alignment is completed, the three types of time series are structured and integrated into a unified dataset structure. Each time axis scale position corresponds to the relevant data content of the three types of time series, and finally a multi-dimensional time series experimental dataset containing state changes, behavior changes, and decision changes is obtained.

[0032] In this embodiment of the invention, by performing timestamp verification and ascending sorting on the associated record data, a continuous experimental process data chain is constructed, overcoming the technical problems of disordered storage of traditional experimental record data and difficulty in continuously tracing back the experimental process in chronological order. This achieves the time-ordered organization of experimental data and ensures the traceability of data throughout the entire experimental process. By classifying and extracting state, behavior, and decision-related data and constructing independent time series for each, the technical problems of mixed storage of traditional experimental data and difficulty in analyzing the changing patterns of data in each dimension are solved, achieving refined time-series sorting of experimental data in each dimension. By calculating and determining the range, scale interval, and number of scales of a unified time axis and completing the time axis alignment of multiple sequences, the technical defects of traditional experimental data lacking a unified time benchmark and difficulty in time-series correlation analysis of various types of data are overcome, achieving time synchronization and integration of experimental state, behavior, and decision data. The finally constructed multi-dimensional time-series experimental dataset provides a structured and standardized data analysis foundation for subsequent time-series correlation feature analysis of the experimental process and automatic identification of key nodes and abnormal events, enhancing the analytical value of experimental data and solving the core problems of fragmented traditional experimental data and inability to perform in-depth time-series analysis.

[0033] In a preferred embodiment of the present invention, step 3 above may include: Step 3.2.1: Extract all data related to experimental state parameters from the experimental process data chain and arrange them in timestamp order to obtain the time series of experimental state parameter change trajectories; extract all data related to experimental behavior execution from the experimental process data chain and arrange them in timestamp order to form the time series of experimental behavior execution timing; extract all data related to experimental decision triggering from the experimental process data chain and arrange them in timestamp order to form the time series of experimental decision triggering nodes. Specifically, this includes: performing a full traversal of the experimental process data chain, accurately classifying and filtering all data nodes in the data chain according to the preset classification rules of data attributes; the first category is filtering data related to experimental state parameters, extracting all state-related data such as image / video analysis parameters of the experimental object, microscopic imaging feature parameters, temperature and humidity / air pressure / ventilation volume parameters of the experimental environment, and rotation speed / temperature / pressure / flow rate of the experimental device, while extracting the unified format timestamp bound to each data point. After identifying the experimental task and removing incomplete or invalid timestamps, the filtered valid experimental state parameter data are rearranged in ascending order of timestamps from earliest to latest, forming a time series of experimental state parameter change trajectories. If the experiment monitors N types of core state parameters, an independent parameter data column is established for each type of parameter in the time series, with the number of data entries in each column matching the number of valid records for that type of parameter. The second category of filtering involves data related to the execution of experimental behaviors. All behavioral data, including experimental execution actions, parameter adjustment actions, response actions, and experimental intervention actions, are extracted from the data chain. Simultaneously, the action type, execution parameters, action result, binding timestamp, and experimental task identifier of each behavioral data entry are extracted. After removing invalid data, all valid experimental behavior execution data are arranged in ascending order of timestamps, forming a time series of experimental behavior execution sequences. Each data node in this sequence uniquely corresponds to one experimental behavior, and the total number of data nodes equals the total number of valid experimental behaviors during the experiment.

[0034] The third category involves filtering experimental decision-triggered data, extracting all decision-related data from the data chain, including experimental decision instructions, decision adjustments, path selection, and path switching. For each decision data point, the system extracts the decision type, decision content, selection criteria, binding timestamp, and experimental task identifier. After removing invalid data, all valid experimental decision-triggered data are sorted in ascending order of timestamp, forming a time series of experimental decision-triggered nodes. Each data node in this series uniquely corresponds to one experimental decision, and the total number of data nodes equals the total number of valid experimental decisions during the experiment. After constructing the three types of time series, the system assigns a unique sequence identifier to each type of sequence and retains the association index between each data point in each sequence and the corresponding data node in the experimental process data chain, ensuring that the time series data can be traced back to the original data chain.

[0035] Step 3.2.2: Integrate and align the time series of experimental state parameter change trajectories, the time series of experimental behavior execution sequences, and the time series of experimental decision trigger nodes according to a unified time axis to generate a multidimensional time series experimental dataset. Specifically, this includes: integrating and aligning the three types of time series according to a unified time axis to generate a multidimensional time series experimental dataset; constructing a unified time axis; extracting all valid timestamps from the three types of time series; determining the start and end times of the time axis, where the start time is the earliest time point among all timestamps and the end time is the latest time point among all timestamps; and calculating the total duration of the time axis, which is the number of milliseconds from the end time minus the number of milliseconds from the start time, in milliseconds. Based on the high-frequency acquisition standard preset for the experiment, the scale interval of the unified time axis is determined. The scale interval is calculated by dividing 1 by the experimental acquisition frequency in seconds. To convert it to milliseconds, the scale interval is multiplied by 1000. If the experimental acquisition frequency is 2 Hz (2 data points per second), the scale interval is 1 ÷ 2 = 0.5 seconds, corresponding to 500 milliseconds. If the acquisition frequency is 1 Hz, the scale interval is 1 second, corresponding to 1000 milliseconds. Based on the start time, end time, and scale interval, the total number of scales on the unified time axis is calculated. The total number of scales is the total duration of the time axis divided by the scale interval. If the total duration of the time axis is 30,000 milliseconds and the scale interval is 500 milliseconds, the total number of scales is 30,000 ÷ 500 = 60. According to this calculation result, the system generates all subsequent scale points sequentially according to the scale interval, with the start time as the first scale point, forming a unified time axis covering the entire experimental process. Each scale point is marked with a unique millisecond-level timestamp.

[0036] Alignment of three types of time series with a unified time axis is performed. The experimental state parameter change trajectory, experimental behavior execution sequence, and experimental decision trigger node are all mapped onto this unified time axis. For time axis tick points with data in each of the three types of series, the corresponding data is completely filled into the corresponding data area of ​​that tick point, labeled as the state parameter area, behavior execution area, and decision trigger area, respectively. For areas with no corresponding data at a time axis tick point, they are marked as null values, and the tick point's position is retained without data loss removal, ensuring the continuity and integrity of the time axis. If a tick point has data in all three types simultaneously... According to the data, the corresponding content is filled into the three data areas at the time point to realize the synchronous presentation of multi-dimensional data at a single time point. After completing the mapping and alignment of all data, all data on the unified time axis are encapsulated in a structured manner to establish the association mapping relationship between the time axis time point and the three types of data. At the same time, the original attributes, association indexes and experimental task identifiers of each type of data are retained. The encapsulated structured data set is defined as a multi-dimensional time series experimental dataset. This dataset contains the independent change trajectories of experimental states, behaviors and decisions arranged in chronological order, and also realizes the linkage presentation of the three types of data under a unified time benchmark.

[0037] In this embodiment of the invention, by classifying and filtering the experimental process data chain and constructing three independent time series according to timestamps, the technical problems of traditional experimental data mixed storage, intertwined and messy data of various dimensions, and difficulty in analyzing state changes, behavior execution and decision triggering patterns separately are overcome. This achieves refined and time-series organization of experimental data of various dimensions, facilitating the individual mining of the change characteristics of each type of data. By calculating and determining the core parameters of a unified time axis and aligning the time axes of the three types of sequences, the technical defects of traditional experimental data lacking a unified time benchmark, the time sequence of state-behavior-decision data being disconnected, and the inability to perform multi-dimensional linkage analysis are compensated for. This achieves synchronous integration of the three types of data at the same time scale, establishing the experimental state... This study establishes a foundation for the temporal correlation of state changes, behavioral execution, and decision triggering. By marking null values ​​at data-free points on the timeline instead of removing them, the continuity of the timeline is ensured, resolving the problem of temporal discontinuity in traditional experimental records caused by data gaps. This ensures the integrity of the temporal dimension of the entire experimental process. The resulting multidimensional time-series experimental dataset provides a structured, standardized, and interconnected data analysis foundation for the temporal correlation feature analysis of the experimental process and the automatic identification of key nodes and abnormal events. This enhances the analytical depth and utilization value of experimental data, and solves the core problems of fragmented traditional experimental records and the inability to analyze deep correlations, influence relationships, or temporal response relationships from the data structure level.

[0038] In a preferred embodiment of the present invention, step 4 above may include: Step 4.1: Extract time series data from the multidimensional time series experimental dataset and analyze the temporal correlation features between the change trajectory of experimental state parameters, the execution sequence of experimental behaviors, and the triggering nodes of experimental decisions. Specifically, this includes: accurately extracting three independent types of time series data from the multidimensional time series experimental dataset according to the sequence identifier: the change trajectory of experimental state parameters, the execution sequence of experimental behaviors, and the triggering nodes of experimental decisions. At the same time, retrieve the unified timestamp, experimental task identifier, and original data association index bound to each data in each type of sequence. Preprocess the three types of time series data, and eliminate random noise in the data acquisition process using the moving average method. The size of the moving window is set according to the experimental acquisition frequency, and the calculation method is to multiply the acquisition frequency by 2. If the acquisition frequency is 2 Hz, that is, 2 data are acquired per second, the moving window size is 2×2=4. If the acquisition frequency is 1 Hz, the moving window size is 2, to ensure the authenticity of the data change trend. After preprocessing, the system initiates time-series correlation feature analysis. Using a unified timeline scale as a benchmark, it matches the time correspondence between experimental decision triggering, experimental behavior execution, and changes in experimental state parameters one by one. It calculates the time difference between the timestamp of each decision triggering / behavior execution and the timestamp of subsequent significant changes in state parameters. The time difference is the number of milliseconds of the state parameter change timestamp minus the number of milliseconds of the decision / behavior trigger timestamp. Simultaneously, it calculates the change in state parameters within a preset time range before and after the decision triggering / behavior execution. The change is the state parameter value after the decision / behavior execution minus the state parameter value before execution. The time difference and parameter change are correlated and matched. Then, by comparing the natural change trend of state parameters without decision / behavior intervention, the system determines the time-series correlation features between the experimental state parameter change trajectory and the experimental behavior execution timeline and experimental decision triggering nodes. These features include the time lag of the correlation trigger, the magnitude correlation of parameter changes, and the driving force of behavior / decision on state changes. The system stores all the analyzed time-series correlation features in a structured manner and establishes associations with the corresponding time nodes and data segments.

[0039] Step 4.2: Based on the aforementioned temporal correlation features, identify the moments when experimental state parameters change or deviate from the expected trend, and mark these moments as key nodes; based on the temporal correlation features, identify the abnormal fluctuation ranges of the experimental state caused by experimental behavior or decision triggers, and mark these ranges as abnormal events. Specifically, this includes: after completing the temporal correlation feature analysis, based on the stored temporal correlation features, automatically identifying key nodes and abnormal events in the experimental process, retrieving the pre-set expected trend indicators of state parameters, including the normal range of parameter changes, reasonable rate of change, stable fluctuation threshold, and other core thresholds; comparing the actual parameter values ​​and actual rate of change in the experimental state parameter change trajectory time series with the preset thresholds one by one, calculating the deviation between the actual parameter values ​​and the expected parameter values. The deviation is the absolute value of the actual parameter value minus the expected parameter value divided by the expected parameter value. If the deviation of the state parameter at a certain time point exceeds the preset deviation threshold, or the rate of change of the state parameter exceeds the preset rate, the deviation is considered. A threshold is used to determine when an experimental state parameter changes or deviates from the expected trend. The timestamp of this moment, the corresponding state parameter value, and the deviation feature are marked as key nodes in the experimental process. Based on temporal correlation features, the range of state parameter changes corresponding to each experimental action or decision trigger is located. The fluctuation amplitude of the state parameter within this range is calculated, which is the maximum value minus the minimum value within the range. If the fluctuation amplitude exceeds the preset abnormal fluctuation threshold, and the fluctuation is directly caused by the corresponding action or decision trigger, the range is determined to be an abnormal fluctuation range of the experimental state. The duration of this range is calculated, which is the number of milliseconds from the end timestamp of the range minus the start timestamp. The start and end timestamps of the range, the fluctuation features, and the corresponding action / decision trigger information are completely marked as abnormal events in the experimental process. All key nodes and abnormal events are bound to the corresponding experimental task identifier and original data index.

[0040] Step 4.3, based on the marked key nodes and abnormal events, analyze the correlation, influence, or temporal response relationships between state changes, behavior execution, and decision triggering during the experiment. Determine the final state, stage output results, or comprehensive results of the experiment as the experimental outcome. Specifically, this includes: sorting all key nodes and abnormal events in ascending order of timestamps, systematically reviewing the experimental behavior execution and experimental decision triggering information corresponding to each key node and abnormal event, and establishing a causal link from experimental decision triggering to experimental behavior execution to changes in experimental state parameters to the generation of key nodes / abnormal events, based on the temporal correlation features obtained in Step 4.1. Identify the direct triggering factors and indirect influencing factors for each key node and abnormal event; for events without behavior / Key nodes that arise naturally under decision-making intervention are analyzed separately for their correlation, influence, or temporal response relationships with basic conditions such as the experimental environment and device operation. After completing the full causal link analysis, the correlation, influence, or temporal response relationships throughout the entire experimental process are integrated into an analysis network using the timeline as a guide. The complete trajectory of the experimental state parameters from initial to final values ​​is tracked. Combined with the final output data of the experimental device and the final state performance of the experimental object, the final state or output result of the experiment is comprehensively determined and taken as the core experimental result. At the same time, the experimental result is linked and bound to the correlation, influence, or temporal response relationship analysis network to ensure that the experimental result can be traced back to the state changes, behavior execution, and decision triggering throughout the entire process.

[0041] Step 4.4: Extract feature parameters from the experimental results. These feature parameters should include at least the state values, change rates, durations of anomalous events, and their impact range at key nodes, to obtain experimental result features. Specifically, this includes: extracting feature parameters from the determined experimental results to form experimental result features; extracting the actual values ​​of experimental state parameters at all marked key nodes as key node state value features; then calculating the change rate of state parameters at key nodes. The change rate is calculated by dividing the change in state parameters between the current key node and the previous adjacent key node by the time interval between the two nodes, i.e., change rate = (state value of the next key node - state value of the previous key node) ÷ (timestamp of the next key node in milliseconds - timestamp of the previous key node in milliseconds). The result is the parameter change per unit millisecond, which is used as the change rate feature; then extracting the values ​​of all marked anomalous events... The duration, calculated time difference of the abnormal interval, and the impact range of the abnormal event are all calculated. The impact range includes the number of experimental state parameters affected by the abnormal event, and the degree to which each affected parameter deviates from the expected trend. The degree is the maximum deviation of the parameter from the expected value within the abnormal interval, which serves as the duration and impact range characteristics of the abnormal event. The above feature parameters are the core content of the experimental result features. The system integrates all extracted feature parameters in a structured manner, labeling each feature parameter with the corresponding experimental task identifier, key node / abnormal event index, and calculation basis, ultimately forming a complete experimental result feature. All feature parameters are stored in a dedicated experimental result feature database to provide data support for the automatic generation of experimental reports. Experimental result features may include one or more of the following: key node features, abnormal event features, state evolution features, behavioral response features, and decision response features.

[0042] In this embodiment of the invention, the technical means of analyzing the temporal correlation features of three types of sequences in a multidimensional time series dataset after preprocessing, automatically identifying key nodes and anomalous events based on preset thresholds and temporal correlation features and completing quantitative calculations, sorting out the causal links corresponding to key nodes and anomalous events to determine experimental results, and accurately calculating and extracting multidimensional experimental result feature parameters according to standardized formulas, effectively overcome the limitations of traditional experimental analysis in uncovering the temporal correlation features between state, behavior, and decision data; the subjectivity and easy omissions and misjudgments in manual identification of key nodes and anomalous events; and the correlation, influence, or temporal response between experimental behavior and state changes. This invention addresses the technical challenges of fuzzy and untraceable relationship analysis, and the lack of standardized calculations and low accuracy in extracting experimental results features. It achieves deep temporal correlation analysis of multi-dimensional data throughout the experimental process; automatically and accurately identifies key experimental nodes and abnormal events to improve the objectivity of analysis; clarifies the correlation, influence, or temporal response relationships between experimental state changes, behavioral execution, and decision triggers to form an interpretable experimental process analysis; and standardizes the extraction of multi-dimensional experimental result feature parameters to provide accurate and structured data support for subsequent automatic report generation, thereby enhancing the depth, scientific rigor, and traceability of experimental result analysis.

[0043] In a preferred embodiment of the present invention, step 5 above may include: Step 5.1: Based on the characteristics of the experimental results, construct the framework of the experimental report. The framework includes an experimental condition summary, an overview of the experimental process, a list of key nodes, a results analysis section, and a conclusion section. Specifically, this includes: constructing a standardized experimental report framework based on the characteristics of the experimental results; retrieving the stored structured data of the experimental results characteristics; identifying the type of this experiment, core monitoring parameters, and key analysis dimensions; matching the standardized report writing specifications of the corresponding experimental field; and building a basic framework for the experimental report that includes an experimental condition summary, an overview of the experimental process, a list of key nodes, a results analysis section, and a conclusion section. Fixed data filling rules and content presentation formats are set for each module within the framework. Specifically, the experimental condition summary uses a table format of parameter items + numerical values ​​+ units; the overview of the experimental process uses a chronological textual description format; the list of key nodes uses an item format of serial numbers + core information; the report content can be presented in the form of text, tables, charts, sequence summaries, index lists, or combinations thereof; the results analysis section uses a textual description format of causal analysis + data support; and the conclusion section uses a summary description format of results + problems + patterns. At the same time, each framework module is configured with a dedicated data retrieval index, and the index is precisely associated with the multidimensional time series experimental dataset, the experimental result feature database, and the key node and abnormal event labeling library.

[0044] Step 5.2: Extract experimental starting conditions and environmental parameters from the multidimensional time series experimental dataset and populate them into the experimental condition summary; extract the main behavioral sequences and state change trajectories during the experimental execution process from the multidimensional time series experimental dataset and generate an experimental process overview. Specifically, this includes: extracting data to populate the experimental condition summary and generating an experimental process overview. In the experimental condition summary population step, the system uses a preset data retrieval index to extract all initial state parameters at the start of the experiment from the multidimensional time series experimental dataset, including initial attribute parameters of the experimental object, initial temperature, humidity, and air pressure parameters of the experimental environment, and initial rotation speed and pressure parameters of the experimental device. Simultaneously, it extracts environmental constraints and device operation constraints that remain unchanged during the experiment. All extracted parameters are categorized and organized according to a fixed format of parameter category-parameter name-initial value-unit of measurement-constraint range, sorted from high to low priority according to the preset core parameters of the experiment, and then completely populated into the preset table of the experimental condition summary. In the experimental process overview generation step, the system... First, the system extracts all experimental behavior execution data from the multidimensional time series experimental dataset. Duplicate routine operations are removed according to preset rules, retaining only key experimental behaviors such as parameter adjustment, decision path switching, and response actions. The number of key behaviors extracted is the total number of experimental behaviors multiplied by a preset screening ratio, which is set to 30% to 50% depending on the experimental type. For example, if the total number of experimental behaviors is 80 and the preset screening ratio is 40%, then 32 key experimental behaviors will be extracted. Next, the system extracts the complete change trajectory of all core state parameters preset for the experiment from the start to the end. The extracted key experimental behaviors are precisely correlated with the changes in core state parameters within the corresponding time range in ascending order of timestamps. The system then narrates the complete process of experiment start-up, execution of each key behavior, corresponding changes in core state parameters, and experiment termination in chronological text, generating an experimental process overview. The level of detail in the narration dynamically adjusts with the number of extracted key behaviors; the more key behaviors, the more concise the narration of each behavior. The above is just an example.

[0045] Step 5.3: Organize the key nodes and abnormal events into a key node list in chronological order, and associate each node with a corresponding state data fragment and behavior description. Specifically, this includes: organizing the key node list and associating each node with a corresponding state data fragment and behavior description. The system first retrieves all key nodes and abnormal event information marked in step 4. After merging the two types of information, it sorts them uniformly in ascending order of timestamp from earliest to latest. Each sorted piece of information is assigned a unique sequence number, forming an initial list of key nodes. The list is then configured with four basic fields: sequence number, timestamp, node or event type, and core features. The corresponding information is then fully populated into each basic field. The core feature field is filled with core analysis data such as parameter deviation of key nodes and parameter fluctuation amplitude of abnormal events. Each node or event in the list is associated with a corresponding state data segment. The data segment retrieval range is the content of a multi-dimensional time series dataset whose timestamp is shifted forward by a preset time to shifted backward by a preset time. The offset time is calculated based on the experimental acquisition frequency, multiplied by 1, in seconds. If the experimental acquisition frequency is 2 Hz, the offset time is 1 second, corresponding to retrieving 2 state data segments before and after the node or event; if the acquisition frequency is 1 Hz, the offset time is 1 second, corresponding to retrieving 1 state data segment before and after the node or event. Meanwhile, the system extracts a complete description of the experimental behavior corresponding to the node or event from the timing data of the experimental behavior execution, including the behavior type, execution parameters, triggering basis, etc. After supplementing the original data association index of the state data fragment and the complete behavior description to the corresponding entries in the key node list, a complete key node list is formed, ensuring that each entry in the list can be directly traced back to the original experimental data through the index.

[0046] Step 5.4: Based on the experimental results characteristics, analyze the correlation, influence, or temporal response relationships between state changes and behavioral decisions during the experiment, and generate a results analysis text. Integrate the experimental process and results analysis to generate experimental conclusions, and combine all parts into a structured experimental report. Specifically, this includes: analyzing correlations, influences, or temporal response relationships to generate results analysis text; extracting experimental conclusions and integrating them into a structured experimental report. In the results analysis text generation stage, the system analyzes the network based on core parameters such as key node state values, parameter change rates, duration of abnormal events, and scope of impact from the experimental results characteristics, combined with the identified correlations, influences, or temporal response relationships of the experimental process, to analyze the experimental state changes. The system conducts a systematic quantitative analysis of the correlation, influence, and temporal response relationships between experimental behaviors and decisions. For each key node and abnormal event, it clarifies the directly triggering behavioral or decision factors, the specific change patterns of the corresponding state parameters, and the degree of influence of the behavior or decision on the experimental state changes. The degree of influence is determined by calculating the ratio of the parameter change after the behavior or decision is executed to the parameter value of the final experimental result, i.e., influence = parameter change after the behavior or decision is triggered ÷ parameter value of the final experimental result. All analysis results are presented in a structured manner according to the time sequence of nodes and events, generating a results analysis text. Each analysis conclusion in the text is labeled with the corresponding experimental result characteristic parameters and original data index to ensure that the analysis conclusions are supported by accurate data. In the stage of extracting experimental conclusions, the system integrates the full-process information of the experimental process overview and the causal analysis conclusions of the results analysis text to extract the core achievements of this experiment, including the final state achieved by the experiment, the core output results, and the verified experimental laws. At the same time, it summarizes the abnormal problems that occurred during the experiment and their clear causes. Finally, based on the core experimental results and the problems found, it proposes optimization directions or relevant experimental suggestions for future experiments, integrating these contents into concise and clear experimental conclusions. After generating the content for each module, the system will fill in the completed experimental condition summary, the generated experimental process overview, the complete list of key nodes, the result analysis text, and the experimental conclusions. It will then modularly integrate these according to the preset framework structure, standardize the format and optimize the layout of each module's content, and add basic information such as experimental task identifiers, experimental start and end times, and data collection frequency to the report. Finally, it will generate a complete structured experimental report. All data-related content in the report will retain the original experimental data's correlation index, enabling traceable referencing between the report content and the original experimental data.

[0047] In this embodiment of the invention, the technical means of constructing a standardized experimental report framework based on experimental result characteristics and experimental domain standards, accurately extracting data from multi-dimensional time-series experimental datasets according to preset rules to fill experimental conditions and associate key behaviors and state trajectories to generate an overview of the experimental process, sorting key nodes and abnormal events by time and associating them with traceable state data fragments and behavioral descriptions, quantitatively analyzing the correlation, influence, or time-series response relationships between state changes and behavioral decisions based on experimental result characteristics, and comprehensively extracting experimental conclusions from the entire process information, and finally integrating all modules to generate a structured experimental report, effectively overcomes the shortcomings of traditional experimental report generation, such as reliance on manual organization of scattered data, lack of a unified standard for report frameworks, lack of data support for experimental process descriptions, incomplete records of key nodes and abnormal events without original data backtracking, strong subjectivity in result analysis and ambiguity in analyzing correlation, influence, or time-series response relationships, and fragmented report content that makes it difficult to support experimental reproduction. The technical challenges of auditing have led to the standardization and normalization of experimental report frameworks, ensuring a consistent and readable report structure across different experiments. It automatically extracts experimental data and fills in report content, reducing the workload and time cost of manual report writing and improving report generation efficiency. It provides traceable original data and behavioral descriptions linking key nodes and abnormal events, making the report content more rigorous and verifiable. Based on quantified experimental result characteristics, it analyzes correlations, influence relationships, or time-series response relationships, making experimental result analysis more scientific and objective, avoiding subjective biases from manual analysis. The resulting structured experimental report integrates core information from the entire experimental process and achieves precise correlation between the report and the original experimental data. This provides complete and traceable report support for experimental reproduction, result auditing, and the accumulation and reuse of experimental knowledge. Experimental conclusions can include one or more of the following: core results, abnormal situations, pattern summaries, and optimization suggestions.

[0048] like Figure 2 As shown, embodiments of the present invention also provide a system for high temporal resolution recording of experimental processes and automatic generation of experimental reports, comprising: The acquisition module is used to continuously collect multi-source state data during the experiment at a frequency of seconds, sub-seconds, or higher. The multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision and path selection data. A timestamp is added to each piece of multi-source state data. The association module is used to associate the collected multi-source state data with the corresponding experimental behaviors, experimental intervention actions and experimental decisions, and to build a mapping relationship between experimental states and behaviors in order to obtain associated record data. The building module is used to organize the associated record data in chronological order and build a multidimensional time series experimental dataset that includes state changes, behavior changes, and decision changes; The identification module is used to analyze multidimensional time series experimental datasets, identify key nodes, key changes or abnormal events in the experimental process, and extract experimental result features. The processing module is used to automatically generate a structured description of experimental results and an experimental report based on the characteristics of the experimental results. The experimental report includes at least the experimental conditions, experimental process, key nodes, result analysis and conclusions, and establishes a correlation and reference relationship with the experimental process data, key node data and / or abnormal event data.

[0049] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0050] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0051] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0052] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for high temporal resolution recording of experimental processes and automatic generation of experimental reports, characterized in that, The method includes: Multi-source state data during the experiment is continuously collected at a frequency of seconds, sub-seconds or higher. The multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision and path selection data. A timestamp is added to each piece of multi-source state data. The collected multi-source state data is associated with the corresponding experimental behaviors, experimental intervention actions and experimental decisions to construct a mapping relationship between experimental states and behaviors, so as to obtain associated record data. Organize the associated record data in chronological order to construct a multidimensional time series experimental dataset that includes state changes, behavior changes, and decision changes; Analyze multidimensional time series experimental datasets to identify key nodes, key changes, or anomalous events in the experimental process, and extract features of the experimental results; Based on the characteristics of the experimental results, a structured description of the experimental results and an experimental report are automatically generated. The experimental report includes at least the experimental conditions, experimental process, key nodes, result analysis and conclusions, and establishes a correlation and reference relationship with the experimental process data, key node data and / or abnormal event data.

2. The method for high temporal resolution recording and automatic experimental report generation of the experimental process according to claim 1, characterized in that, Multi-source state data is continuously acquired during the experiment at a frequency of seconds, sub-seconds, or higher. This multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision-making and path selection data. A timestamp is added to each piece of multi-source state data, including: Images, videos, or microscopic imaging data of the experimental object are continuously acquired at a frequency of seconds, sub-seconds, or higher using an image sensor, and a timestamp is added to each data point; different types of data, such as experimental environment or device operation, can be acquired using the same or different sampling frequencies, aligned using a unified time reference, and a time stamp is established for each data point; experimental execution action data are synchronously recorded through experimental execution control, and a timestamp is added to each data point; Key result information is obtained from external experimental equipment or modules. The acquisition methods include display content recognition, interface reading, log parsing, data synchronization, or a combination thereof, and a timestamp is added to each piece of information. Experimental decision and path selection data are recorded synchronously through experimental decision-making, and a timestamp is added to each piece of data to obtain multi-source state data with timestamps.

3. The method for high temporal resolution recording and automatic experimental report generation of the experimental process according to claim 2, characterized in that, The collected multi-source state data is correlated with corresponding experimental behaviors, intervention actions, and decisions to construct a mapping relationship between experimental states and behaviors, resulting in associated record data, including: A unique experimental task identifier is generated for each experimental process; all collected multi-source state data, as well as experimental behaviors, intervention actions, and decision-making data during the experimental process, are associated with the experimental task identifier. Record the control behaviors, experimental intervention actions or parameter adjustment information, parameter values ​​before and after adjustment and the basis for adjustment related to changes in experimental state during the experiment, and associate the parameter adjustment instructions, parameter values ​​before and after adjustment and the basis for adjustment with the corresponding state data; Record the state data, execution parameters, and response actions before and after the occurrence of the abnormal event, and associate the state data, execution parameters, and response actions before and after the occurrence of the abnormal event with the experimental task identifier to obtain associated record data containing the mapping relationship between state and behavior.

4. The method for high temporal resolution recording and automatic experimental report generation of the experimental process according to claim 3, characterized in that, The associated record data is organized chronologically to construct a multidimensional time-series experimental dataset containing state changes, behavioral changes, and decision changes, including: The associated record data is arranged in chronological order according to timestamps to form a continuous and traceable experimental process data chain; Based on the experimental process data chain, multiple time series are constructed, corresponding to the change trajectory of experimental state parameters, the execution sequence of experimental behavior, and the triggering node of experimental decision, respectively, to obtain a multi-dimensional time series experimental dataset.

5. The method for high temporal resolution recording and automatic experimental report generation of the experimental process according to claim 4, characterized in that, Based on the experimental process data chain, multiple time series are constructed, corresponding to the change trajectory of experimental state parameters, the execution sequence of experimental behaviors, and the triggering nodes of experimental decisions, respectively, to obtain a multi-dimensional time series experimental dataset, including: Extract all data related to experimental state parameters from the experimental process data chain and arrange them in timestamp order to obtain the time series of experimental state parameter change trajectories; extract all data related to experimental behavior execution from the experimental process data chain and arrange them in timestamp order to form the time series of experimental behavior execution timing; extract all data related to experimental decision triggering from the experimental process data chain and arrange them in timestamp order to form the time series of experimental decision triggering nodes. The time series of experimental state parameter change trajectories, experimental behavior execution time sequences, and experimental decision triggering nodes are integrated and aligned according to a unified time axis to generate a multi-dimensional time series experimental dataset. The multi-dimensional time series experimental dataset supports backtracking to the original experimental process data.

6. The method for high temporal resolution recording and automatic experimental report generation of the experimental process according to claim 5, characterized in that, Analyze multidimensional time series experimental datasets to identify key nodes, critical changes, or anomalous events during the experimental process, and extract features from the experimental results, including: Extract time series data from a multidimensional time series experimental dataset, and analyze the change trajectory of experimental state parameters, the execution sequence of experimental behaviors, and the temporal correlation characteristics between the triggering nodes of experimental decisions; Based on the aforementioned temporal correlation features, the moments when experimental state parameters change or deviate from the expected trend are identified and marked as key nodes; based on the temporal correlation features, the intervals of abnormal fluctuations in experimental state caused by experimental behavior or decision triggers are identified and marked as abnormal events. Based on the marked key nodes and abnormal events, the correlation, influence, or temporal response relationships between state changes, behavior execution, and decision triggering during the experiment are analyzed to determine the final state, stage output results, or comprehensive results of the experiment, which are then used as the experimental results. Feature parameters are extracted from the experimental results. These feature parameters include at least the state values ​​at key nodes, the rate of change, the duration of abnormal events, and the scope of their impact, in order to obtain the characteristics of the experimental results.

7. The method for high temporal resolution recording and automatic experimental report generation of the experimental process according to claim 6, characterized in that, Based on the characteristics of the experimental results, a structured description of the experimental results and an experimental report are automatically generated. The experimental report includes at least the experimental conditions, experimental process, key milestones, result analysis, and conclusions, and establishes a correlation and reference relationship with the experimental process data, key milestone data, and / or abnormal event data, including: Based on the characteristics of the experimental results, a framework for the experimental report is constructed, which includes a summary of experimental conditions, an overview of the experimental process, a list of key nodes, a results analysis section, and a conclusion section. Extract the experimental starting conditions and environmental parameters from the multidimensional time series experimental dataset and populate them into the experimental condition summary; extract the main behavioral sequences and state change trajectories during the experimental execution process from the multidimensional time series experimental dataset to generate an overview of the experimental process; Organize key nodes and abnormal events into a list of key nodes in chronological order, and associate each node with a corresponding state data fragment and behavior description; Based on the characteristics of the experimental results, the correlation, influence, or time-series response relationships between state changes and behavioral decisions during the experiment are analyzed, generating a results analysis text. Integrating the experimental process and results analysis, experimental conclusions are generated, and all parts are combined into a structured experimental report. A structured description of the experimental results and the experimental report are automatically generated. The experimental report includes one or more of the following: text, tables, charts, time-series summaries, key node lists, or combinations thereof. At least a portion of the content in the experimental report establishes an indexed reference relationship with the original experimental data, key node data, or anomalous event data.

8. A system for high temporal resolution recording of experimental processes and automatic generation of experimental reports, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The acquisition module is used to continuously collect multi-source state data during the experiment at a frequency of seconds, sub-seconds, or higher. The multi-source state data includes image or video data of the experimental object, experimental environment or device operation data, experimental execution action data, and experimental decision and path selection data. A timestamp is added to each piece of multi-source state data. The association module is used to associate the collected multi-source state data with the corresponding experimental behaviors, experimental intervention actions and experimental decisions, and to build a mapping relationship between experimental states and behaviors in order to obtain associated record data. The building module is used to organize the associated record data in chronological order and build a multidimensional time series experimental dataset that includes state changes, behavior changes, and decision changes; The identification module is used to analyze multidimensional time series experimental datasets, identify key nodes, key changes or abnormal events in the experimental process, and extract experimental result features. The processing module is used to automatically generate a structured description of experimental results and an experimental report based on the characteristics of the experimental results. The experimental report includes at least the experimental conditions, experimental process, key nodes, result analysis and conclusions, and establishes a correlation and reference relationship with the experimental process data, key node data and / or abnormal event data.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images