Laboratory energy-saving control system and method

Abstract

The invention discloses a laboratory energy-saving control system and method, and belongs to the field of equipment energy-saving control. The laboratory energy-saving control system comprises a power utilization evaluation module, a time analysis module and a task scheduling module; the power utilization evaluation module collects power utilization parameters of equipment, calculates a fluctuation intensity coefficient and a trend change index, analyzes the correlation degree, reduces dimensions to obtain a power utilization evaluation value, and dynamically adjusts the collection frequency according to the power utilization evaluation value; the time analysis module constructs a time sequence incidence matrix, fuses the time-space characteristics to obtain weighted characteristics, clusters and divides operation stages, and constructs an equipment time-energy consumption thermodynamic diagram; the task scheduling module ranks the priority according to the remaining time length of the task, screens candidate time periods, calculates and distributes an optimal value in combination with energy consumption, electricity price and the like, selects an execution time period and updates thermodynamic diagram data; accurate acquisition and analysis of power consumption data, accurate division of equipment operation stages and energy consumption visualization are realized, task scheduling is optimized, laboratory energy consumption is reduced, and resource utilization efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of equipment energy-saving control, and particularly relates to a laboratory energy-saving control system and method. BACKGROUND

[0002] During the operation of laboratory equipment (such as precision instruments, constant temperature equipment, etc.), there are problems of complex power consumption state, significant energy consumption fluctuation, and lack of precise basis for task scheduling: On the one hand, the equipment operation includes different stages such as temperature rise, steady state, standby, etc., and the power consumption parameters (power, current, etc.) fluctuate greatly in different stages, but the traditional monitoring adopts fixed frequency data collection, which easily leads to insufficient data precision in high fluctuation stage and resource waste in stable stage; On the other hand, laboratory tasks are diverse and have different execution time requirements, and existing scheduling mostly relies on manual experience without dynamic optimization combining equipment energy consumption state and power grid price, which often results in situations such as allocating high-power tasks in high energy consumption period (such as equipment temperature rise period) and emergency tasks being overdue due to time period conflict, not only causing energy waste but also affecting the reliability of experimental process; In addition, when multiple devices run in parallel, there is a lack of intuitive energy consumption space-time distribution visualization tool, and management personnel cannot quickly locate high energy consumption links, which cannot provide data support for energy-saving strategy making, resulting in high overall energy consumption of the laboratory and low resource utilization efficiency, for which we propose a laboratory energy-saving control system and method. SUMMARY

[0003] The purpose of the present application is to provide a laboratory energy-saving control system and method to solve the problems raised in the background.

[0004] To achieve the above purpose, the present application provides the following technical solution: a laboratory energy-saving control system, comprising: An electricity evaluation module: collects the power consumption parameters of the equipment, sets a time window, calculates the fluctuation intensity coefficient and trend change index of the power consumption parameter data in the time window, and performs comprehensive analysis to obtain the fluctuation characteristic value, constructs the power consumption characteristic matrix, analyzes the correlation degree of the power consumption parameters and the overall energy consumption fluctuation and marks them positively and negatively, and after standardization and dimension reduction, calculates the electricity evaluation value, and accordingly matches the collection frequency of the power consumption parameters in the time window; A time analysis module: constructs a time sequence correlation matrix, analyzes the weighted characteristics of the time window, collects the historical average weighted characteristics of each time window of the equipment as samples, divides the running stage clusters, merges the continuous time windows into running stage time slices, and constructs the equipment time-energy heat map; A task scheduling module: receives the heat map data, obtains the task queue execution time, calculates the remaining time length of the task, and accordingly sorts the priority, filters the candidate allocation period, calculates the allocation optimization value, and accordingly selects the task execution period.

[0005] Preferably, the specific process of constructing the electricity feature matrix is as follows: For each device, the electricity parameter data is synchronously collected and recorded, and the electricity parameter data includes: instantaneous power, current, voltage, rated power; A time window is set, and for each item of electricity parameter data, the variance value of the time window is obtained, which is denoted as the fluctuation intensity coefficient; The time series and parameter values of the electricity parameter data in the time window are fitted to obtain a time trend line, and the absolute value of the slope of the time trend line is obtained, which is denoted as the trend change index; After the fluctuation intensity coefficient and the trend change index corresponding to the electricity parameter data are assigned different preset weight coefficients, the weighted sum is performed to obtain the fluctuation feature value; The electricity parameter data in the time window and the corresponding fluctuation feature value are combined to form the electricity feature matrix.

[0006] Preferably, the specific process of calculating the electricity evaluation value is as follows: For each item of electricity parameter data, the correlation quantitative value of the electricity parameter data and the overall energy consumption fluctuation of the device is analyzed and marked, and the specific process is as follows: The fluctuation feature value sequence of the electricity parameter and the numerical sequence of the overall energy consumption fluctuation of the device in each continuous time window in the historical data are extracted; The total number and proportion of forward synchronous changes and reverse changes of the fluctuation feature value sequence of the electricity parameter and the numerical sequence of the overall energy consumption fluctuation of the device are counted, and the one with a larger proportion is selected as the correlation quantitative value and marked as the direction; The fluctuation feature values of the electricity parameters in the electricity feature matrix are standardized, the parameters are divided into positive and negative correlation categories according to the correlation positive and negative, the parameters in each category are sorted according to the absolute value of the correlation quantitative value, and the first n parameters in each category are selected to reduce the dimension of the matrix; The correlation quantitative values of the 2n dimensions after dimension reduction are normalized to obtain the weight of each dimension, the numerical values of each dimension are extracted, the direction coefficient is combined, and the electricity evaluation value is calculated.

[0007] Preferably, the specific process of matching the collection frequency of the time window electricity parameter is as follows: A collection frequency adjustment strategy library is established, the strategy library sets multiple electricity evaluation value intervals, and each electricity evaluation value interval is pre-set to correspond to a target collection frequency; The electricity evaluation value calculated in the current time window is substituted into the collection frequency adjustment strategy library, the corresponding target collection frequency is output, and the target collection frequency is used as the actual collection frequency of the electricity data in the next time window.

[0008] Preferably, the specific process of analyzing the weighted features of the time window is as follows: It receives electricity consumption parameter data and application evaluation values ​​for each time window in real time, and integrates the electricity consumption parameters and derived features of each time window into a multi-dimensional feature vector. The multidimensional feature vector contains: the mean values ​​of instantaneous power, voltage, and current within the time window, as well as the corresponding fluctuation intensity coefficient, trend change index, and electricity consumption assessment value; Construct a temporal correlation matrix from the eigenvectors of a continuous time window; The cosine similarity algorithm is used to calculate the similarity between the current time window and each historical time window. The similarity is normalized to obtain the temporal attention weight of each historical window, and temporal weighted features are generated accordingly. Extract the correlation metric values ​​between each dimension of the multidimensional feature vector and the overall energy consumption fluctuation, normalize them, obtain the spatial attention weights, and then generate spatial weighted features. By assigning different weight coefficients to the time and space weighted features and summing them up, we obtain the current time window weighted features containing multi-dimensional information.

[0009] Preferably, the average weighted features of each time window within the daily operating cycle of the device in the past week are collected and used as the weighted features of the corresponding time window in the current operating cycle, with each time window corresponding to one sample; For each sample, label the power and electricity consumption assessment values ​​for the corresponding time window; Cluster all samples within the current running cycle to obtain clusters for different running stages; For each cluster of operating stages, based on the sample power and power consumption assessment value, the category of operating stage to which it belongs is defined, including: heating operation stage, steady-state operation stage, and standby operation stage.

[0010] Preferably, the specific process for constructing the equipment time-energy consumption heatmap is as follows: Construct a two-dimensional Cartesian coordinate system, with the horizontal axis representing the current operating cycle duration of the device and the vertical axis representing the device ID; For each time slice of the operation phase, obtain the average energy consumption and average power consumption assessment value of the equipment within that time slice; Three base colors are preset, each corresponding to a different stage of operation: For each operating phase, multiple average energy consumption intervals and average electricity consumption assessment intervals are divided, and a corresponding reference color brightness value is matched for each average energy consumption interval, and a corresponding reference color saturation value is matched for each average electricity consumption assessment interval. The average energy consumption and power consumption assessment mean of each time slice in the coordinate system are matched with the average energy consumption interval and power consumption assessment mean interval of the corresponding operating stage to determine the corresponding brightness and saturation values. Finally, the matched color parameters are displayed in the coordinate area corresponding to the time slice to form a device time-energy consumption heat map.

[0011] Preferably, the specific process of calculating the remaining time of tasks and prioritizing them accordingly is as follows: Receive device time-energy consumption heatmap data generated by the time analysis module; Get the current task queue of the device and mark the task execution time of each task in the task queue; For any task in the task queue, obtain the historical average completion time of the task. The remaining time for task scheduling is obtained by subtracting the time from the current time when the task was executed from the corresponding historical average completion time. Based on the remaining time of each task in ascending order, the tasks are dynamically prioritized to obtain a task priority sequence.

[0012] Preferably, the specific process for selecting the task execution time period is as follows: Runtime segments are allocated to each task in sequence according to the task priority sequence; For tasks to be assigned, starting from the current time and ending at the task execution time, extract all time windows with a continuous duration greater than or equal to the historical average completion time of the task, excluding the warming cluster period, to form a preliminary set of selectable time periods; The selectable time period is divided into multiple sub-time periods of equal length by a sliding window based on the average completion time of the task in history. Each sub-time period corresponds to a candidate allocation time period. For any candidate allocation period, the average energy consumption, the average electricity assessment value, and the real-time electricity price of the power grid corresponding to that period are obtained, and after normalization, the optimal allocation value is obtained. Select the time period with the highest allocation preference value from all candidate allocation time periods as the target execution time period for the task; Each time a task is assigned, the core characteristic parameters such as the average energy consumption and electricity consumption assessment value of the execution period in the equipment time-energy consumption heatmap are updated based on the historical average energy consumption of the task. The above steps are repeated to complete the assignment of all tasks in the task queue. The equipment time-energy consumption heatmap is updated daily based on the daily operation cycle duration.

[0013] Compared with the prior art, the beneficial effects of the present invention are: (1) The laboratory energy-saving control system and method collects multi-dimensional power consumption parameters through the power consumption assessment module, combines time window analysis to calculate the fluctuation intensity coefficient and trend change index, screens core energy consumption related parameters and dynamically adjusts the acquisition frequency—increasing the acquisition density during the high fluctuation stage to ensure data accuracy, and reducing the frequency during the stable stage to save resources. This provides a high-quality data foundation for subsequent analysis and avoids the resource waste and accuracy imbalance problems of traditional fixed frequency acquisition.

[0014] (2) The laboratory energy-saving control system and method integrates historical and real-time power consumption characteristics through a time analysis module using a spatiotemporal attention mechanism, and accurately divides the equipment into heating, steady state and standby stages using a density clustering algorithm. At the same time, it constructs a time-energy consumption heat map of the equipment and uses color mapping to intuitively present the spatiotemporal energy consumption differences of multiple devices. Managers can quickly locate high-energy consumption periods and low-energy consumption links, providing clear data support for the formulation of energy-saving strategies and solving the problems of traditional technologies being unable to dynamically identify operating stages and lacking energy consumption visualization tools.

[0015] (3) The laboratory energy-saving control system and method dynamically sorts the priority of tasks based on the remaining time of the task through the task scheduling module to ensure that emergency tasks are executed first; the time period without high energy consumption heating clusters is selected as candidates, and the optimal value is calculated and the time period is allocated by combining energy consumption, electricity price and parameter matching degree to achieve dual energy saving of "off-peak electricity use + stable state"; at the same time, the heat map data is updated after the task is executed to form a scheduling closed loop, which not only avoids the energy waste of allocating high-power tasks during high energy consumption periods, but also ensures the reliability of task execution, and solves the problem that traditional manual scheduling lacks data support and it is difficult to balance energy saving and task execution. Attached Figure Description

[0016] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0018] Please see Figure 1 The present invention provides a laboratory energy-saving control system and method, including: an electricity consumption assessment module, a time analysis module and a task scheduling module; The electricity consumption assessment module collects the electricity consumption parameters of the equipment, sets a time window, calculates the fluctuation intensity coefficient and trend change index of the electricity consumption parameter data within the time window, performs comprehensive analysis to obtain fluctuation characteristic values, constructs a matrix, analyzes the correlation between electricity consumption parameters and overall energy consumption fluctuations and marks them positive and negative, standardizes and reduces dimensionality, calculates the electricity consumption assessment value, and matches the collection frequency of electricity consumption parameters within the time window accordingly. The specific process is as follows: Smart meters are installed on the power supply lines of each piece of equipment in the laboratory; power consumption parameter data is collected and recorded synchronously for each piece of equipment, including instantaneous power, current, voltage, rated power, etc. Set a time window; for each power consumption parameter data, obtain the variance value of the time window, which is recorded as the fluctuation intensity coefficient; used to quantify the degree of data dispersion and characterize the severity of fluctuations in the power consumption status of equipment; Meanwhile, based on the least squares method, the time series of power consumption parameter data within the time window is fitted with the parameter values ​​to obtain a time change trend line. The absolute value of the slope of the time change trend line is obtained and recorded as the trend change index; this is used to measure the rate of change of the equipment's power consumption parameters within the time period. After assigning different preset weight coefficients to the fluctuation intensity coefficient and trend change index corresponding to the electricity consumption parameter data, the fluctuation characteristic value is obtained by weighted summation. The electricity consumption parameter data within the time window are combined with their corresponding fluctuation characteristic values ​​to form an electricity consumption characteristic matrix; For each power consumption parameter, its correlation with the overall energy consumption fluctuation of the equipment is analyzed and quantified, and then labeled as positive or negative. The specific process is as follows: Extract the fluctuation characteristic value sequence of electricity consumption parameters in each continuous time window and the numerical sequence of overall energy consumption fluctuation of the equipment from historical data; If the fluctuation characteristic value of the electricity consumption parameter data shows an upward trend and the overall energy consumption fluctuation also shows an upward trend, or if the fluctuation characteristic value shows a downward trend and the overall energy consumption fluctuation also shows a downward trend, it is recorded as a positive synchronous change. The total number of such positive synchronous changes is counted, and the proportion of these changes to the total number of counts is calculated and denoted as the positive correlation frequency ratio. If the fluctuation characteristic value increases but the overall energy consumption fluctuation decreases, or the fluctuation characteristic value decreases but the overall energy consumption fluctuation increases, it is recorded as an inverse change. The total number of such reverse changes is counted, and its proportion of the total number of counts is calculated and denoted as the negative correlation frequency ratio. Compare the frequency percentages of positive associations with those of negative associations: If the proportion of positive association frequency is greater than the proportion of negative association frequency, the proportion of positive association frequency is taken as the association metric and marked as positive association degree; If the proportion of negative association frequency is greater than the proportion of positive association frequency, the proportion of negative association frequency is taken as the association metric and marked as negative association degree; If the two are the same, mark them as having zero correlation. The larger the absolute value of the correlation quantification value, the closer the correlation between the electricity consumption parameter and the overall energy consumption fluctuation.

[0019] Data standardization processing is performed on the fluctuation characteristic values ​​corresponding to different electricity consumption parameters in the electricity consumption characteristic matrix; For each power consumption parameter data in the power consumption feature matrix, based on the positive or negative relationship between each power consumption parameter data and the correlation quantification value of the overall energy consumption fluctuation of the equipment, they are divided into positive correlation class and negative correlation class. In the positive correlation class and the negative correlation class, each power consumption parameter data is sorted according to the magnitude of the absolute value of its corresponding correlation quantification value, and the top n power consumption parameter data are selected from each sequence to reduce the dimensionality of the power consumption feature matrix, so as to obtain the dimensionality data of the top n power consumption parameter data retained after dimensionality reduction of the positive correlation class and the dimensionality data of the top n power consumption parameter data retained after dimensionality reduction of the negative correlation class. Normalize the absolute values ​​of the correlation metrics corresponding to the 2n dimensions after dimensionality reduction to obtain the weights wi for each dimension; extract the values ​​xi for each dimension after dimensionality reduction. Using the formula: The electricity consumption assessment value S is obtained; Where i is the index of each dimension after dimensionality reduction, and its value ranges from 1 to 2n; n represents the number of the first n preset electricity consumption parameter data selected in each of the positive and negative correlation classes; Sgni is the directional coefficient corresponding to the i-th dimension. If the correlation between the power consumption parameter data corresponding to this dimension and the overall energy consumption fluctuation of the equipment is marked as a positive correlation, then Sgni=1; if it is marked as a negative correlation, then Sgni=-1, which is used to reflect the direction of the correlation. The higher the electricity consumption assessment value, the more significant the fluctuation in the equipment's electricity consumption status or the closer it is to the overall energy consumption. Establish a collection frequency adjustment strategy library. This strategy library sets multiple power consumption assessment value ranges, and each power consumption assessment value range is preset with a corresponding target collection frequency. The electricity consumption assessment value calculated in the current time window is substituted into the acquisition frequency adjustment strategy library, the corresponding target acquisition frequency is output, and the target acquisition frequency is used as the actual acquisition frequency of electricity consumption data in the next time window.

[0020] It should be noted that by collecting instantaneous power, current and other parameters in real time through smart meters, and combining the variance of the time window (fluctuation intensity coefficient) and the least squares fitting slope (trend change index), a quantitative analysis of the dispersion and rate of change of electricity consumption data can be achieved. This can accurately capture the subtle fluctuations in the power consumption status of equipment and lay a data foundation for subsequent energy consumption analysis. By using correlation analysis to screen out core parameters closely related to overall energy consumption fluctuations, eliminating redundant dimensions, and then calculating the power consumption assessment value through weighted calculation after dimensionality reduction, not only is the data processing efficiency improved, but the correlation between equipment power consumption status and overall energy consumption can also be accurately reflected, providing data support for energy-saving strategies. The data collection frequency is automatically adjusted based on the power consumption assessment value. When the power consumption fluctuates drastically (such as during the heating phase), the collection density is increased to ensure data accuracy. When the power consumption is stable (such as during the steady-state phase), the frequency is reduced to save system resources and achieve a balance between data quality and resource consumption. It provides standardized power consumption feature vectors for the time analysis module, helping to accurately divide the operation phase (heating / steady state / standby); at the same time, it provides energy consumption status references for the task scheduling module, enabling task allocation to avoid high energy consumption periods (such as heating clusters), and optimizes the scheduling strategy by combining electricity price and parameter matching degree, ultimately achieving a reduction in overall laboratory energy consumption and efficient utilization of resources.

[0021] Time Analysis Module: Constructs a time-series correlation matrix, analyzes the weighted features of time windows, collects historical average weighted features of each time window of the equipment as samples, divides the operation phase into clusters, merges continuous time windows into operation phase time slices, and constructs a time-energy consumption heatmap of the equipment. The specific process is as follows: Real-time reception of electricity consumption parameter data and corresponding electricity consumption assessment values ​​for each time window; For each time window, the electricity consumption parameters and derived features within that time window are integrated into a multi-dimensional feature vector. ; t is the label of the current time window; The multidimensional feature vector contains: the mean values ​​of instantaneous power, voltage, and current within the time window, as well as the corresponding fluctuation intensity coefficient, trend change index, and electricity consumption assessment value; For continuous Feature vectors for each time window: A temporal correlation matrix is ​​constructed to capture feature correlations over time; where The preset number of historical windows; The cosine similarity algorithm is used to calculate the similarity between the current time window and each historical time window. The calculation formula is: Get the current window With historical window Similarity of features Where k is the offset of the historical window relative to the current window, used to identify different historical time windows. A multidimensional feature vector for a historical time window with an offset of k; The similarity between the current time window and each historical time window is normalized to obtain the time attention weight of each historical time window. ; The calculation formula is: ; in, This is the exponential transformation value of the similarity between the current time window and a historical time window with an offset of k, used to amplify the difference in similarity. For the current time window and all The sum of the similarity index transformation values ​​of each historical time window, where m is the index of the historical time window offset; Based on the obtained time attention weights for each historical time window, time-weighted features are generated. The formula is: ; The significance of this formula lies in the fact that the feature vectors of each historical time window are weighted and summed according to the corresponding time attention weights, and then added to the feature vector of the current time window. This not only preserves the features of the current time window, but also integrates the parts of the features of each historical time window that are relevant to the current analysis, thus achieving an effective fusion of the current state and historical trends. For each dimension of the multidimensional feature vector, extract the correlation metric values ​​between each dimension and the overall energy consumption fluctuation of the device; normalize these correlations to obtain the spatial attention weights of each dimension; the larger the weight, the stronger the influence of the corresponding dimension on energy consumption fluctuations. Based on the spatial attention weights of each dimension, the multidimensional feature vector is weighted and summed in each dimension to generate spatially weighted features; Different weight coefficients are assigned to time-weighted features and spatial-weighted features, and then a weighted sum is performed to obtain the current time window weighted features containing multi-dimensional information. The weighted features are a set of multi-dimensional features that integrate time and spatial dimensions. For example, if the original feature vector contains 6 dimensions such as power and voltage, then the time-weighted features and spatial-weighted features are both 6-dimensional vectors, and the fused weighted features are also 6-dimensional vectors. Each dimension aggregates historical trends and parameter importance information through spatiotemporal weights. The temporal attention weighting aggregates historical trends and current states, while the spatial attention weighting focuses on key parameter dimensions. This allows for precise characterization of equipment power consumption characteristics within the current time window, supporting core functions such as subsequent operational phase division and energy consumption trend prediction. It avoids the limitations of single-dimensional feature analysis by deeply integrating the temporal correlation of equipment power consumption with parameter importance, laying a data foundation for phase identification, heat map construction, and task scheduling optimization in laboratory energy-saving control, and realizing the key transformation from raw power consumption data to energy-saving decisions. Collect the average weighted features of each time window within the daily operating cycle of the equipment over the past week, and use them as the weighted features of the corresponding time window in the current operating cycle. Each time window corresponds to one sample. Label each sample with the power and electricity consumption assessment values ​​for the corresponding time window. Among them, the power mean is the original average value of the instantaneous power within the time window in the weighted feature (calculated based on the average value of the same time window within the most recent week). Electricity consumption assessment value: This is a comprehensive assessment value calculated based on the original parameters in the weighted features (calculated based on the average value of the same time window within the most recent week). The specific process of clustering all samples within the current running cycle using the density clustering algorithm to obtain clusters for different running stages is as follows: Determine the key clustering parameters, namely the neighborhood radius and the minimum number of samples. The specific process for determining the cluster neighborhood radius is as follows: calculate the Euclidean distance between all samples in the current running cycle, statistically analyze the distance distribution, and take the median multiplied by a preset adjustment coefficient as the neighborhood radius; the minimum number of samples is preset based on the characteristics of a week's data to ensure effective clustering of similar samples. After setting the parameters, the determined neighborhood radius and minimum number of samples are substituted into the density clustering algorithm to iteratively cluster all samples in the current running cycle. The algorithm scans the sample space and divides density-connected sample points into the same cluster: if a sample point contains more than or equal to the minimum number of samples in its neighborhood, it is marked as a core point, and all directly density-reachable sample points in its neighborhood are merged into a cluster; for sample points that cannot be assigned to any cluster, they are marked as noise points; after the above clustering process, several running stage clusters are finally obtained. For each operational phase cluster, based on the sample power and electricity consumption assessment values, its operational phase category is defined, and the specific process is as follows: If the samples in the operating phase cluster meet the following conditions: the slope of the fitted line of the power of all samples in the cluster is greater than the preset upward trend threshold, and the mean of the power consumption assessment value is greater than the preset heating power consumption assessment value threshold, then the operating phase cluster is defined as the heating cluster. If the variance of the power of all samples in the operating phase cluster is less than the preset steady-state power fluctuation threshold, and the variance of the electricity consumption assessment value is less than the preset steady-state electricity consumption assessment value fluctuation threshold, then the operating phase cluster is defined as a steady-state cluster. If the average power of all samples in the operating phase cluster is less than the preset standby power ratio threshold (this threshold is set according to a certain proportion of the device's rated power), and the average power consumption assessment value is less than the preset standby power consumption assessment value threshold, then the operating phase cluster is defined as a standby cluster. Based on the time sequence, consecutive time windows belonging to the same cluster within the current operating cycle of the equipment are merged into a continuous operating phase. Accordingly, the duration of the current operating cycle of the equipment is divided into several operating phase time slices to avoid misjudgment of phases caused by fluctuations in a single window. Among them, the heating cluster corresponds to the equipment heating stage, the steady-state cluster corresponds to the equipment steady-state stage, and the standby cluster corresponds to the equipment standby stage. The specific process for constructing a device time-energy consumption heatmap is as follows: Construct a two-dimensional Cartesian coordinate system, with the horizontal axis representing the current operating cycle duration of the equipment (divided according to the time slice interval of the operating phase) and the vertical axis representing the equipment ID; form a two-dimensional grid of "equipment - operating phase time slice" to intuitively present the spatiotemporal energy consumption distribution of multiple equipment; For each time slice of the operation phase, obtain the average energy consumption and average power consumption assessment value of the equipment within that time slice; Three preset reference colors are C1, C2, and C3. C1 corresponds to the heating-up operation stage, C2 corresponds to the steady-state operation stage, and C3 corresponds to the standby operation stage. For each operating phase, multiple average energy consumption intervals and average electricity consumption assessment intervals are divided, and a corresponding reference color brightness value is matched for each average energy consumption interval, and a corresponding reference color saturation value is matched for each average electricity consumption assessment interval. The average energy consumption and power consumption assessment mean of each time slice in the coordinate system are matched with the average energy consumption range and power consumption assessment mean range of the corresponding operating stage to determine the corresponding brightness and saturation values. Finally, the matched color parameters (base color + brightness + saturation) are displayed in the coordinate area corresponding to the time slice to form a device time-energy consumption heat map.

[0022] It should be noted that the average power consumption parameters, fluctuation intensity, trend indicators, and power consumption assessment values ​​of the time window are integrated into a multi-dimensional feature vector. By combining cosine similarity and spatiotemporal attention mechanism, the historical trend and current state are fused through time-weighted features, and key parameter dimensions (such as power and voltage) are focused on by spatial attention weights. This avoids the limitations of single-dimensional analysis and accurately captures the temporal correlation and parameter importance of equipment power consumption characteristics. Based on the density clustering algorithm, by dynamically calculating the neighborhood radius (adjusted by the median distance of historical samples) and the preset minimum number of samples, the equipment operation cycle is divided into clusters such as heating, steady state, and standby. Combined with the power slope, variance and evaluation value threshold to define the stage category, the single window fluctuation interference is effectively eliminated, and the dynamic identification of the equipment operation status is realized, providing a time period basis for energy consumption optimization. Construct a device time-energy consumption heatmap, using a two-dimensional coordinate system combined with color mapping (base color brightness / saturation corresponds to energy consumption and evaluation value range), to intuitively present the spatiotemporal energy consumption distribution of multiple devices at different operating stages, which facilitates the rapid location of low energy consumption periods (such as steady-state clusters) and high energy consumption periods (such as heating clusters), providing a visual decision reference for time period optimization in the task scheduling module; The system provides equipment operation phase division and energy consumption heat map data to the task scheduling module, enabling task allocation to avoid high-energy-consumption heating clusters. It also optimizes scheduling strategies by combining electricity prices and parameter matching. At the same time, by rolling updates of historical average weighted features (data from the most recent week), the system ensures that the model dynamically adapts to equipment energy consumption patterns, improves the real-time performance and accuracy of energy-saving control, and achieves closed-loop optimization from data feature extraction to energy-saving decision-making.

[0023] The task scheduling module receives heatmap data, obtains the marked execution times of the task queue, calculates the remaining time of the tasks, sorts the priorities accordingly, filters candidate allocation time slots, calculates the optimal allocation value, and selects the task execution time slot based on this. The specific process is as follows: Receive device time-energy consumption heatmap data generated by the time analysis module; Get the current task queue of the device and mark the task execution time of each task in the task queue; For any task in the task queue, obtain the historical average completion time of the task. The remaining time for task scheduling is obtained by subtracting the time from the current time when the task was executed from the corresponding historical average completion time. Based on the remaining time of task scheduling, tasks are dynamically prioritized in ascending order to obtain a task priority sequence. Runtime segments are allocated to each task in sequence according to the task priority sequence; For tasks to be assigned, starting from the current time and ending at the task execution time, extract all time windows with a continuous duration greater than or equal to the historical average completion time of the task, excluding the warming cluster period, to form a preliminary set of selectable time periods; The selectable time period is divided into multiple sub-time periods of equal length by a sliding window based on the average completion time of the task in history. Each sub-time period corresponds to a candidate allocation time period. For any candidate allocation period, obtain the average energy consumption, the average electricity assessment value, and the real-time electricity price of the grid corresponding to that period, and after normalization, use the formula: The optimal allocation value YXZ is obtained; Where α, β, γ, and δ are preset weighting coefficients; Emin is the lowest historical average energy consumption of the device, and Ecurr is the average energy consumption during the candidate allocation period; Smean is the mean of the electricity consumption assessment values ​​for the candidate allocation period (after normalization, the value is between 0 and 1). Cmin is the lowest real-time electricity price of the power grid, and Ccurr is the average real-time electricity price during the candidate allocation period; M represents the degree of matching between the task parameters and the equipment operation phase; Select the time period with the highest allocation preference value from all candidate allocation time periods as the target execution time period for the task; Each time a task assignment is completed, the core characteristic parameters of the execution period in the device time-energy consumption heatmap are updated based on the historical average energy consumption of the task. The core characteristic parameters include: the average energy consumption of the target execution period, the electricity consumption assessment value, etc. Repeat the above steps to complete the assignment of all tasks in the task queue in sequence; The equipment time-energy consumption heatmap is updated daily based on the duration of a single day's operation cycle.

[0024] The analysis process for the matching degree between task parameters and equipment operation phases is as follows: Obtain the required range of each execution parameter for the task. Execution parameters include data such as execution power, voltage, and current. Obtain the range of values ​​for the task execution parameters in the candidate allocation time period; For each execution parameter of the task, its required execution range is compared with the candidate allocation time period parameter range, and the matching sub-value of that parameter is calculated, specifically: If the execution range of the task parameters is completely within the range of the candidate allocation time period parameters, then the matching sub-value is 1; If the execution range of the task parameters is within the range of the candidate allocation time period parameters, obtain the length of the overlapping interval between the execution range of the task parameters and the range of the candidate allocation time period parameters, and the total length of the execution range of the task parameters. By dividing the length of the overlapping interval by the total length of the execution range of the task parameters, the matching sub-value is obtained. If the parameter execution range is completely outside the range of the candidate allocation time period parameters, then the matching sub-value is 0; Using the formula: The matching degree M between the task parameters and the equipment operation stage is obtained, where g is the label of the task execution parameter, H is the total number of task execution parameters, mg is the matching sub-value corresponding to the g-th execution parameter, and Dg is the preset weight coefficient corresponding to the g-th execution parameter. The process of updating the core feature parameters can be as follows: For any core feature parameter, the updated parameter value = (original parameter value × historical statistics count + task impact value × preset impact coefficient) / (historical statistics count + 1) Among them, the historical statistics count is the cumulative number of times the parameters have been updated during this execution period; The task impact value is the actual data generated by this task execution on this parameter (e.g., when calculating average energy consumption, it is the historical average energy consumption of this task; when calculating electricity consumption assessment value, it is the actual assessment value of this task).

[0025] It should be noted that the remaining time is calculated by "time elapsed since task execution - historical average completion time". Tasks are prioritized from shortest to longest to ensure that urgent tasks (such as tasks with insufficient remaining time) are allocated time slots first, avoid task timeouts, and ensure the continuity of the experimental process. Update the task queue status in real time to adapt to dynamic task scenarios in the laboratory (such as temporarily added tasks) and improve the system's response flexibility; Select time periods without heating clusters as candidates to avoid high power fluctuations and energy consumption during the equipment heating phase, thereby reducing overall energy consumption from the perspective of time period selection; Combining energy consumption (compared to historical minimum values), electricity consumption assessment values ​​(state stability), real-time electricity prices (economic efficiency), and task matching degree, the optimal value is calculated through a weighted formula, taking into account both energy-saving efficiency and task feasibility; for example, the time periods of low electricity prices (low Ccurr) and equipment steady-state periods (low Smean) are prioritized for allocation, achieving dual energy saving of "low-off-peak electricity consumption + state stability". The matching degree is calculated by the overlap rate between the task execution parameters (power, voltage, etc.) and the parameter range of the candidate time period. This ensures that the task requirements are compatible with the equipment operation phase (e.g., high-power tasks are matched with steady-state cluster time periods), reduces equipment abnormalities or energy waste caused by parameter mismatch, and improves equipment utilization and task execution reliability. After each task assignment, the core features of the heatmap (such as the average energy consumption of the target period and the electricity consumption assessment value) are updated based on the historical average energy consumption of the task. This enables subsequent scheduling to be optimized based on the latest energy consumption data, forming a closed loop of "scheduling-update-rescheduling" and gradually improving the model's prediction accuracy of equipment energy consumption patterns. Daily data reset: Heat map data is updated based on the daily operation cycle to adapt to the typical scenario of "daily start-up and shutdown" of laboratory equipment and ensure that the scheduling strategy is in line with the real-time operation status; This module achieves the following through the logic of "priority sorting → candidate time period filtering → multi-factor optimization → data closed-loop update": Energy efficiency improvement: Avoid scheduling during periods of high energy consumption and off-peak electricity prices to reduce the overall electricity cost of the laboratory; Task execution reliability: Parameter matching analysis ensures smooth task execution and reduces equipment wear and tear; System Adaptability: Dynamically updates heat map data and continuously optimizes scheduling strategies to adapt to scenarios such as equipment aging and changes in task types, providing a feasible technical path for intelligent energy-saving control in laboratories.

[0026] A laboratory energy-saving control method, comprising: Step 1: Collect the power consumption parameters of the equipment, set a time window, calculate the fluctuation intensity coefficient and trend change index of the power consumption parameter data within the time window, and conduct comprehensive analysis to obtain the fluctuation characteristic value. Construct a power consumption characteristic matrix, analyze the correlation between power consumption parameters and overall energy consumption fluctuation and mark them as positive and negative, standardize and reduce the dimensionality, calculate the power consumption assessment value, and match the collection frequency of power consumption parameters in the time window accordingly. Step 2: Construct a time-series correlation matrix, analyze the weighted features of time windows, collect historical average weighted features of each time window of the equipment as samples, divide the operation phase clusters, merge continuous time windows into operation phase time slices, and construct the equipment time-energy consumption heat map. Step 3: Receive heatmap data, obtain the task queue marked execution time, calculate the remaining time of the task, sort the priorities accordingly, filter candidate allocation time periods, calculate the preferred allocation value, and select the task execution time period accordingly.

[0027] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A laboratory energy saving control system characterized by: Comprise: Power consumption evaluation module: collect power consumption parameters of the equipment, set time window, calculate fluctuation intensity coefficient and trend change index of power consumption parameter data in the time window, and conduct comprehensive analysis to obtain fluctuation characteristic value, construct power consumption characteristic matrix, analyze the correlation degree of power consumption parameters and overall energy consumption fluctuation and mark positive and negative, standardize and reduce dimension, calculate power consumption evaluation value, and match the collection frequency of power consumption parameters in the time window according to the power consumption evaluation value; Time analysis module: construct time sequence correlation matrix, analyze weighted characteristics of time window, collect historical average weighted characteristics of each time window of the equipment as samples, divide running stage clusters, merge continuous time windows into running stage time slices, and construct equipment time-energy heat map; Task scheduling module: receive heat map data, obtain task queue label execution time, calculate task remaining time, and sort priority according to the task remaining time, select candidate allocation period, calculate allocation optimization value, and select task execution period according to the allocation optimization value.

2. The laboratory energy conservation control system of claim 1, wherein: The specific process of constructing the power consumption characteristic matrix is: For each device, the power consumption parameter data is collected synchronously and recorded, and the power consumption parameter data includes: instantaneous power, current, voltage, rated power; Set the time window, for each power consumption parameter data, obtain the variance value of the time window, denoted as the fluctuation intensity coefficient; The time series and parameter value of the power consumption parameter data in the time window are fitted to obtain the time trend line, and the absolute value of the slope of the time trend line is obtained, denoted as the trend change index; After assigning different preset weight coefficients to the fluctuation intensity coefficient and the trend change index corresponding to the power consumption parameter data, weighted sum is performed to obtain the fluctuation characteristic value; The power consumption parameter data in the time window and the corresponding fluctuation characteristic value are combined to form the power consumption characteristic matrix.

3. The laboratory energy conservation control system of claim 2, wherein: The specific process of calculating the power consumption evaluation value is: For each power consumption parameter data, analyze its correlation degree quantization value with the overall energy consumption fluctuation of the equipment and mark positive and negative, the specific process is: Extract the fluctuation characteristic value sequence of the power consumption parameter and the numerical sequence of the overall energy consumption fluctuation of the equipment in each continuous time window in the historical data; Statistical fluctuation characteristic value sequence of power consumption parameters and numerical sequence of overall energy consumption fluctuation of the equipment positive synchronous change and reverse change total number and proportion, select the proportion as the correlation degree quantization value and mark the direction; Standardize the fluctuation characteristic value of the power consumption parameter in the power consumption characteristic matrix, divide the parameter into positive and negative correlation categories according to the correlation degree, and sort the correlation degree quantization value absolute value in each category, and reduce the dimension of the matrix by selecting the first n parameters in each category; Normalize the correlation degree quantization value absolute value of the 2n dimensions after dimension reduction to obtain the weight of each dimension, extract the numerical value of each dimension, combine the direction coefficient, and calculate the power consumption evaluation value.

4. The laboratory energy conservation control system of claim 3, wherein: The specific process of matching the collection frequency of the power consumption parameters in the time window is: Establish a collection frequency adjustment strategy library, which sets multiple power consumption evaluation value intervals in the strategy library, and each power consumption evaluation value interval is pre-set with a corresponding target collection frequency; Put the power consumption evaluation value calculated in the current time window into the collection frequency adjustment strategy library, output the corresponding target collection frequency, and use the target collection frequency as the actual collection frequency of the power consumption data in the next time window.

5. The laboratory energy conservation control system of claim 4, wherein: The specific process of analyzing the weighted features of the time window is as follows: Real-time receiving of the power consumption parameter data of each time window and the power consumption evaluation value thereof, and integration of the power consumption parameter and the derived features of each time window into a multi-dimensional feature vector; The multi-dimensional feature vector contains the mean values of instantaneous power, voltage and current in the time window, corresponding fluctuation intensity coefficients, trend change indicators and power consumption evaluation values; Construction of a time sequence correlation matrix for the feature vectors of consecutive time windows; Calculation of the similarity between the current time window and each historical time window using the cosine similarity algorithm; Normalization of the similarity to obtain the time attention weight of each historical window, and generation of a time weighted feature based thereon; Quantification of the correlation between each dimension of the multi-dimensional feature vector and the overall energy consumption fluctuation, normalization and obtaining of a spatial attention weight, and further generation of a spatial weighted feature; Weighted summation of the time and spatial weighted features with different weight coefficients to obtain a current time window weighted feature containing multi-dimensional information.

6. The laboratory energy-saving control system according to claim 5, characterized in that: Collecting the average weighted features of each time window in the daily operation cycle of the equipment in the last week as the weighted features of the corresponding time window in the current operation cycle, each time window corresponding to a sample; Labeling the power and power consumption evaluation value of the corresponding time window for each sample; Clustering all samples in the current operation cycle to obtain different operation stage clusters; For each operation stage cluster, defining the operation stage category to which it belongs based on the sample power and power consumption evaluation value, including: warming-up operation stage, steady-state operation stage and standby operation stage.

7. A laboratory energy saving control system according to claim 6, wherein: The specific process of constructing the equipment time-energy heat map is as follows: Constructing a two-dimensional rectangular coordinate system, with the horizontal axis representing the current operation cycle length of the equipment and the vertical axis representing the equipment ID; For each operation stage time slice, obtaining the average energy consumption and power consumption evaluation value of the equipment in the time slice; Predefining three reference colors corresponding to different operation stages: For each operation stage, dividing multiple average energy consumption intervals and power consumption evaluation value intervals, and matching the corresponding reference color brightness value for each average energy consumption interval and the corresponding reference color saturation value for each power consumption evaluation value interval; Matching the average energy consumption and power consumption evaluation value of each operation stage time slice in the coordinate system with the average energy consumption interval and power consumption evaluation value interval of the corresponding operation stage to determine the corresponding brightness value and saturation value, and finally displaying the matched color parameters in the coordinate area corresponding to the time slice to form the equipment time-energy heat map.

8. The laboratory energy conservation control system of claim 7, wherein: The specific process of calculating the remaining task duration and sorting the priorities based thereon is as follows: Receiving the equipment time-energy heat map data generated by the time analysis module; Obtaining the task queue of the current equipment and labeling the task execution time of each task in the task queue; For any task in the task queue, obtaining the historical average completion duration of the task, and obtaining the task arrangement remaining duration by subtracting the historical average completion duration from the time duration between the task execution time and the current time; Based on the order of the task arrangement remaining duration from short to long, dynamically sorting the priorities of the tasks to obtain a task priority sequence.

9. A laboratory energy saving control system according to claim 8, wherein: The specific process of selecting the task execution period is as follows: According to the task priority sequence, each task is allocated a running period; For the task to be allocated, a time window with a continuous duration ≥ the historical average completion duration of the task is extracted from the current time as the starting point and the task execution time as the ending point, and a preliminary set of selectable time periods is formed; The selectable time periods are divided into multiple equal-length sub-time periods by sliding window according to the historical average completion duration of the task, and each sub-time period corresponds to a candidate allocation period; For any candidate allocation period, the average energy consumption, electricity evaluation value mean and real-time electricity price of the period are obtained, and after normalization, the allocation optimization value is obtained; The period with the maximum allocation optimization value is selected from all candidate allocation periods as the target execution period of the task; After completing the task allocation once, the average energy consumption, electricity evaluation value and other core characteristic parameters of the execution period in the device time-energy thermal map are updated based on the historical average energy consumption of the task, and the above steps are repeated to complete the allocation of all tasks in the task queue, and the device time-energy thermal map is updated daily based on the single-day operation cycle length.

10. A laboratory energy-saving control method applied to a laboratory energy-saving control system according to any one of claims 1-9, characterized in that: Step one: collect device power parameters, set time window, calculate fluctuation intensity coefficient and trend change index, get fluctuation characteristic value, construct matrix, analyze correlation degree and reduce dimension to calculate evaluation value, and adjust the collection frequency accordingly; Step two: construct time sequence correlation matrix to analyze weighted features, collect historical average features as samples, cluster analysis running phase cluster, merge time slices, and construct device time-energy thermal map; Step three: receive thermal map data, mark task execution time to calculate remaining duration, sort priority, select candidate period to calculate optimization value, and select execution period.

Images