Intelligent scheduling method of construction equipment based on AI algorithm

Abstract

The present invention relates to the technical field of intelligent scheduling of building equipment, and specifically to an intelligent scheduling method for building equipment based on an AI algorithm, comprising the following steps: collecting equipment energy consumption records, user activity data, and meteorological data, performing integrated calculations, and generating comprehensive equipment energy consumption data; using the comprehensive equipment energy consumption data, applying a convolutional neural network to perform model training, and obtaining an equipment prediction model. The present invention integrates the energy consumption records, user activity data, and meteorological data of building equipment, and applies a convolutional neural network to analyze and train models on these data, thereby being able to predict equipment demand in real time and adjust equipment operation. This prediction and adjustment mechanism makes equipment operation more accurate, reduces energy consumption in a targeted manner, and improves efficiency. In addition, continuous monitoring and data feedback loops ensure that models and scheduling strategies are updated in real time, adapt to environmental changes, and optimize equipment performance.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent scheduling of construction equipment, and in particular to an intelligent scheduling method for construction equipment based on an AI algorithm. Background Art

[0002] The field of building equipment intelligent scheduling technology involves the use of information technology and intelligent systems to efficiently manage and automatically schedule various equipment in buildings. The core goal of this technology is to improve energy efficiency, optimize equipment efficiency, and reduce maintenance costs.

[0003] However, existing technologies mainly rely on static procedures and regular inspections to manage and schedule building equipment. For example, without real-time reflection of the environment and equipment status, equipment operation is often not in optimal condition, which leads to poor energy efficiency and waste of resources. In the absence of real-time prediction and automatic adjustment mechanisms, maintenance is usually carried out only after equipment failure occurs. This reactive maintenance strategy not only increases maintenance costs, but may also affect the normal operation of the building. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings existing in the prior art and propose an intelligent scheduling method for construction equipment based on AI algorithm.

[0005] In order to achieve the above object, the present invention adopts the following technical solution, which is an intelligent scheduling method for construction equipment based on AI algorithm, comprising the following steps:

[0006] Collect equipment energy consumption records, user activity data and meteorological data, perform integrated calculations, and generate comprehensive equipment energy consumption data; use the comprehensive equipment energy consumption data to train a model using a convolutional neural network to obtain an equipment prediction model;

[0007] Based on the equipment prediction model, the operation schedule of each equipment in the current environment is calculated to generate a preliminary equipment scheduling plan; the preliminary equipment scheduling plan is simulated and implemented, operation data is collected, and the operation data is analyzed in real time, and equipment operation parameters are adjusted to obtain an optimized equipment scheduling plan;

[0008] Implementing the optimized equipment scheduling plan, monitoring equipment operation and energy consumption in real time, and generating equipment operation effect data; analyzing the equipment operation effect data, comparing it with the output of the equipment prediction model, adjusting model parameters, and obtaining an updated equipment prediction model;

[0009] Based on the updated equipment prediction model, the equipment operation schedule is recalculated periodically.

[0010] Preferably, the steps for acquiring the comprehensive data of equipment energy consumption are:

[0011] Collect energy consumption records of building equipment, user activity data and meteorological data, and perform preprocessing operations on the three types of data, including data format unification and outlier removal, to obtain the preprocessed data set;

[0012] Based on the preprocessed data set, the energy consumption records, user activity data and meteorological data are correlated and analyzed, and the correlation analysis data is obtained by combining the equipment usage and meteorological conditions;

[0013] Based on the correlation analysis data, the data is focused on the factors that affect energy consumption through cluster analysis to generate comprehensive equipment energy consumption data.

[0014] Preferably, the steps of acquiring the device prediction model are:

[0015] Receiving the comprehensive energy consumption data of the equipment, selecting a convolutional neural network, setting the number of network layers, the number of neurons and the activation function, and obtaining a preliminary network model structure;

[0016] Based on the preliminary network model structure, the comprehensive data of device energy consumption is loaded, a model training process is performed, and a trained network model is obtained by iteratively optimizing network weights and biases;

[0017] Verification and testing are performed based on the trained network model, and the prediction ability and accuracy of the model are verified using the test set data to obtain the equipment prediction model.

[0018] Preferably, the steps for obtaining the preliminary equipment scheduling plan are:

[0019] Receiving predicted operation data of each device from the device prediction model, analyzing current environmental variables, including temperature, humidity, and device usage frequency, to obtain environmental and prediction data sets;

[0020] Based on the environment and the predicted data set, the operation scheduling priority of each device is calculated using the following formula:

[0021] ;

[0022] in, For equipment The predicted energy consumption, For equipment The efficiency coefficient, For equipment The efficiency reduction coefficient is For equipment Adjustment factor to match environmental conditions, Schedule priority for operations;

[0023] According to the operation scheduling priority, the equipment operation order is arranged in order of priority, and the interaction impact and energy consumption of the equipment are comprehensively considered to generate a preliminary equipment scheduling plan.

[0024] Preferably, the steps for obtaining the optimized equipment scheduling plan are:

[0025] Implementing the preliminary equipment scheduling plan and performing simulation operation, collecting operation data of equipment performance and environmental response, and obtaining an original operation data set;

[0026] Based on the original running data set, the resource utilization efficiency of the computing device is calculated using the following formula:

[0027] ;

[0028] in, Representative equipment Adjusted resource efficiency, For equipment The actual amount of energy consumed, For simulation prediction of equipment the amount of energy that should be consumed;

[0029] According to the resource utilization efficiency, the equipment operating parameters are adjusted to optimize the overall energy efficiency, thereby obtaining an optimized equipment scheduling plan.

[0030] Preferably, the steps for acquiring the equipment operation effect data are:

[0031] Implement the optimized equipment scheduling scheme, monitor the operating status and energy consumption of each equipment in real time, and obtain a real-time monitoring data set;

[0032] Based on the real-time monitoring data set, the energy efficiency performance index of each device is calculated using the following formula:

[0033] ;

[0034] in, For equipment The energy efficiency performance index, For equipment The total energy consumed during the monitoring period, is the operating output of the equipment during the same period;

[0035] According to the energy efficiency performance index, the operating efficiency of all equipment is summarized and analyzed, and the equipment performance indicators and energy consumption data are integrated to generate equipment operation effect data.

[0036] Preferably, the steps for obtaining the updated device prediction model are:

[0037] Receiving the equipment operation effect data, sorting and classifying the difference items between the equipment operation performance index and the model prediction result, and obtaining the equipment operation difference data;

[0038] Based on the equipment operation difference data, the correlation pattern between the equipment operation and the prediction data is analyzed, the input parameters and structure of the equipment prediction model are adjusted, and an updated model structure is generated.

[0039] Preferably, the step of acquiring the updated device prediction model also includes: retraining and verifying model performance based on the updated model structure, optimizing model weights and bias values ​​in combination with real-time data feedback, and obtaining an updated device prediction model.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are:

[0041] The present invention integrates the energy consumption records of building equipment, user activity data and meteorological data, and applies convolutional neural networks to analyze and train models of these data, so as to predict equipment demand in real time and adjust equipment operation. This prediction and adjustment mechanism makes equipment operation more accurate, reduces energy consumption in a targeted manner, and improves efficiency. In addition, continuous monitoring and data feedback loops ensure that models and scheduling strategies are updated in real time, adapt to environmental changes, and optimize equipment performance. It improves the accuracy and predictability of equipment operation, reduces energy waste, improves environmental quality, and reduces maintenance costs, thereby enhancing overall economic benefits and sustainability. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0044] See also Figure 1 The present invention provides a technical solution, an intelligent scheduling method for construction equipment based on AI algorithm, comprising the following steps:

[0045] Collect equipment energy consumption records, user activity data and meteorological data, perform integrated calculations, and generate comprehensive equipment energy consumption data; use the comprehensive equipment energy consumption data to train the model using a convolutional neural network to obtain an equipment prediction model;

[0046] Based on the equipment prediction model, calculate the operation schedule of each equipment in the current environment and generate a preliminary equipment scheduling plan; simulate and implement the preliminary equipment scheduling plan, collect operation data, and analyze the operation data in real time, adjust the equipment operation parameters, and obtain an optimized equipment scheduling plan;

[0047] Implement optimized equipment scheduling plans, monitor equipment operation and energy consumption in real time, and generate equipment operation effect data; analyze equipment operation effect data, compare it with the output of the equipment prediction model, adjust model parameters, and obtain an updated equipment prediction model;

[0048] Based on the updated equipment prediction model, the equipment operation schedule is recalculated regularly.

[0049] The steps to obtain comprehensive data on equipment energy consumption are as follows:

[0050] Collect energy consumption records of building equipment, user activity data and meteorological data, and perform preprocessing operations on the three types of data, including data format unification and outlier removal, to obtain the preprocessed data set;

[0051] Based on the preprocessed data set, the energy consumption records, user activity data and meteorological data are correlated and analyzed, and the correlation analysis data is obtained by combining the equipment usage and meteorological conditions;

[0052] Based on the correlation analysis data, cluster analysis is used to focus the data on factors that affect energy consumption and generate comprehensive data on equipment energy consumption.

[0053] Specifically, in the process of collecting energy consumption records of building equipment, user activity data and meteorological data, we mainly focus on the power consumption characteristics of the equipment, the periodicity of user activities and the frequency of changes in climate conditions. We perform a series of preprocessing operations on data from different sources. First, each item in the data set is formatted, and the data format is unified into a standard numerical or time series format for subsequent processing. Then, the outliers and noise in the data are removed through the data cleaning process, such as marking and eliminating unconventional high-energy consumption data points in energy consumption records, filling or deleting blank periods in user activity data, and smoothing mutation values ​​in meteorological data. This stage of processing ensures the consistency and accuracy of the data, and provides a high-quality input source for subsequent data analysis. After completing these steps, a preprocessed, clean and uniformly formatted data set is obtained.

[0054] Based on the preprocessed data set, the energy consumption records of building equipment, user activity data and meteorological data are correlated and analyzed. In this process, the relationship between equipment energy consumption and user activity patterns and the impact of meteorological conditions are analyzed through data correlation technology. For example, by calculating the overlap between the time period of high energy consumption of equipment and the peak period of user activity, the direct impact of meteorological conditions such as temperature and humidity on equipment energy consumption is analyzed. Through this multi-dimensional correlation analysis, it can be revealed which factors have the most significant impact on energy consumption. This not only helps to understand the reasons for fluctuations in equipment energy consumption, but also provides scientific data support for equipment management. The result of the correlation analysis is a correlation analysis data set that comprehensively reflects the equipment usage and the impact of meteorological conditions.

[0055] After obtaining the association analysis data set, the data is further focused on the factors that most significantly affect energy consumption through cluster analysis technology. In this step, the key factors of clustering are first determined, such as device type, user activity frequency, and changes in meteorological conditions. Then, clustering algorithms such as K-means are applied to group the data set. Each group represents a specific energy consumption impact pattern. By analyzing the characteristics of the data points in each group, the factors that have the greatest impact on energy consumption can be accurately identified. Such analysis not only improves the efficiency of data processing, but also makes energy consumption management more targeted and effective. What is ultimately generated is a comprehensive device energy consumption data that accurately describes the relationship between each influencing factor and energy consumption.

[0056] The steps to obtain the equipment prediction model are:

[0057] Receive comprehensive data on equipment energy consumption, select a convolutional neural network, set the number of network layers, number of neurons, and activation function, and obtain a preliminary network model structure;

[0058] Based on the preliminary network model structure, the comprehensive data of equipment energy consumption is loaded, the model training process is executed, and the trained network model is obtained by iteratively optimizing the network weights and biases;

[0059] Verification and testing are performed based on the trained network model. The prediction ability and accuracy of the model are verified using the test set data to obtain the equipment prediction model.

[0060] Specifically, after receiving the comprehensive data on equipment energy consumption, first standardize each data type, unify the timestamp format and value range, eliminate outliers in the data set, such as data points with abnormally high or low energy consumption, to ensure data consistency and comparability. This step is key to data preprocessing, making the input data friendly to model training. Then, feature engineering technology is selected to extract effective features from the energy consumption data, such as average energy consumption, peak time, etc., and segment the data into training sets and test sets to ensure that the model can be trained and verified on diverse data and generate a preliminary network model structure.

[0061] Based on the preliminary network model structure, preparatory work before model training is carried out to ensure the efficient transfer of each batch of data between the memory and the processing unit. Batch processing technology is used to ensure the effective management of large amounts of data. Then, forward propagation and back propagation algorithms are implemented for parameter optimization, initial weights and biases are set, and appropriate activation functions, such as ReLU or Sigmoid, are selected to ensure the effectiveness of nonlinear transformations. The cross-entropy loss function is used to regularly evaluate the model error. The training cycle is controlled by setting the number of iterations and batch size. The model is gradually optimized to reduce the risk of overfitting, and the learning rate is adjusted according to real-time feedback to obtain the trained network model.

[0062] According to the trained network model, the model is verified and tested in detail. First, the verification standards are set, such as determining the safety threshold of the model performance, such as the accuracy must be more than 80% and the loss value must be less than 0.05. Then, an independent test set is used to evaluate the model, and key indicators such as accuracy, recall, precision and F1 score are calculated. The performance of the model in different categories of prediction is analyzed, confusion matrix analysis and ROC curve generation are performed, and the reliability and stability of the model in practical applications are evaluated. According to the test results, the model parameters are fine-tuned, such as adjusting the number of layers or modifying the activation function, to improve the overall prediction ability of the model and obtain the device prediction model.

[0063] The steps to obtain the preliminary equipment scheduling plan are:

[0064] Receive the predicted operation data of each device from the device prediction model, analyze the current environmental variables, including temperature, humidity, and device usage frequency, and obtain the environmental and prediction data sets;

[0065] Based on the environment and prediction data set, the operation scheduling priority of each device is calculated using the following formula:

[0066] ;

[0067] in, For equipment The predicted energy consumption, For equipment The efficiency coefficient, For equipment The efficiency reduction coefficient is For equipment Adjustment factor to match environmental conditions, Schedule priority for operations;

[0068] According to the operation scheduling priority, the equipment operation order is arranged in order of priority, and the equipment interaction impact and energy consumption are comprehensively considered to generate a preliminary equipment scheduling plan.

[0069] Specifically, based on the predicted operation data, the predicted operation data of each device is classified and processed, the predicted data of multiple devices in the same time period are integrated into a table form, and the timestamp corresponding to each device is used as the associated key value, the operation characteristics of each device in the same time period are aligned row by row, and the temperature measurement values ​​of multiple locations in the current environment are recorded. A qualified digital temperature sensor is used for multi-point real-time collection, and the multi-point temperature data is transmitted to the data processing terminal through a wired connection. The record is updated every 5 minutes. The humidity data is measured by a digital humidity sensor. The humidity sensor quantifies the humidity signal through an embedded processing chip, and the quantification result is integrated into a percentage value, which is recorded in the data processing terminal memory at the same 5-minute interval. The frequency of use is counted from the equipment operation log statistics. The number of times the equipment is switched on and off within a period and the change in load rate are marked at the corresponding time points. The frequency values ​​are accumulated through the counting program to form a unified data comparison table. In order to set the data fusion standard, the temperature range is distributed between -10℃ and 40℃, and the humidity is between 0% and 100%. The equipment use frequency is counted based on the number of operations per hour, and the high use frequency interval is when the number of operations exceeds 24 times. The above multi-source data is matched and indexed by selecting programming instructions, and the abnormal values ​​in the data are marked and removed to ensure that there are no unmatched row elements in the data table. Finally, the temperature, humidity, equipment use frequency and predicted operation data sets are merged to form a complete environment and prediction data set.

[0070] The benefit of the formula is that by incorporating predicted energy consumption, equipment efficiency attenuation coefficient, environmental condition matching coefficient and efficiency coefficient into the priority calculation, the generated operation scheduling priority can more comprehensively reflect the current actual working status and conditions of the equipment.

[0071] The acquisition steps are as follows: According to the predicted operation data obtained previously, select the equipment within the target time period The predicted energy consumption value is obtained by continuously collecting the equipment power consumption data with power metering instruments and calculating the data through multiple linear regression analysis. ;

[0072] The steps to obtain the efficiency are as follows: according to the equipment benchmark efficiency, the ratio of the equipment's power consumption to its output power under rated conditions is taken as the efficiency coefficient, and the efficiency coefficient is obtained by recording the equipment manual and multiple comparative test results. ;

[0073] The acquisition steps are as follows: according to the equipment life statistics and actual maintenance records, quantify the efficiency attenuation ratio of the equipment in the current period, and compare the rated performance of the equipment with the maintenance test data for many years to obtain ;

[0074] The acquisition steps are as follows: according to the previously obtained environmental and prediction data sets, the temperature, humidity and equipment design conditions are compared and analyzed, and the optimal working conditions of the equipment are compared with the current environmental conditions by querying the chart method, and the equipment adaptability is quantified to obtain .

[0075] Calculation process:

[0076] Substituting the above parameters into the formula:

[0077] ;

[0078] First calculate the subtraction part within the brackets:

[0079] ;

[0080] Then perform the multiplication operation:

[0081] ;

[0082] Continue multiplying by 1.1:

[0083] ;

[0084] Finally divide by 3.5:

[0085] ;

[0086] The result shows that the operation scheduling priority of this device under the current environmental conditions and predicted energy consumption background is about 1.79. This value is greater than 1, which means that under the current conditions, the device is more likely to be arranged higher in the priority sequence than in the baseline state. When the result is higher than 1.5, it can be determined that the device has a higher startup priority, and if it is lower than 0.8, it means that the device can be placed at a later position in the current scheduling sequence.

[0087] According to the operation scheduling priority obtained above, the priority data of each device is read, and each device is sorted with the priority value as the key value. The sorted list is stored in the data recording table. When reading the device interaction impact data, the historical parallel operation of each pair of devices is compared item by item. Whether the two devices are running simultaneously in the same time segment is recorded one by one in the statistical time segment. The counting frequency of simultaneous operation is accumulated as the interaction count value, and the interaction count value is compared with the selected standard threshold. The standard threshold is determined by counting the operation data of multiple months and calculating the average interaction frequency between devices. If the interaction count value is higher than the threshold, it is marked as a high interaction combination in the data recording table. The energy consumption data is retrieved one by one through the historical electricity metering records, and the average power consumption in the same hour segment every day is used as the benchmark value. The benchmark value is recorded in an independent data file, and then the devices are read out from the sorted list one by one, and the energy consumption value of the device in the corresponding time period is compared with the benchmark value. The deviation value exceeding the benchmark value is recorded in a separate column and merged with the results of the marked high interaction combination device. After continuously comparing multiple groups of records, the comprehensive results are stored in the scheduling data file to form a preliminary equipment scheduling plan.

[0088] The steps to obtain the optimized equipment scheduling plan are:

[0089] Implement the preliminary equipment scheduling plan and conduct simulation operation to collect the operation data of equipment performance and environmental response and obtain the original operation data set;

[0090] Based on the original running data set, the resource utilization efficiency of the calculation device is calculated using the following formula:

[0091] ;

[0092] in, Representative equipment Adjusted resource efficiency, For equipment The actual amount of energy consumed, For simulation prediction of equipment the amount of energy that should be consumed;

[0093] According to the resource utilization efficiency, the equipment operating parameters are adjusted to optimize the overall energy efficiency and obtain the optimized equipment scheduling plan.

[0094] Specifically, based on the preliminary equipment scheduling scheme obtained above, the simulation operation result file of the implemented scheduling scheme is parsed, and the power consumption data of each device at each time point is extracted from the file, and the power consumption data is stored in the data table in the form of time series records, and the actual energy consumption value is read from each time period as a row index, and the recorded meteorological sensor data and user activity data are aligned with the power consumption data according to the timestamp. When aligning, the corresponding environmental sensor measurement value and the user activity event number are matched according to the previously marked device location number, and multiple columns of data fields are added to the data table for the corresponding rows of each device and environmental variable, including environmental response values ​​and performance performance parameters, and the environmental response values ​​are standardized through a calibrated scale, and the environmental response values ​​are converted into a numerical range clearly defined in the scale. If missing data occurs during the conversion process, the device self-test log file is queried to supplement the corresponding information, and the user activity event number is converted into the corresponding activity weight by querying the user activity record file, and a new column is added to the data table to store the weight data, and the device performance parameters are accumulated in multiple records and the performance mean is obtained by arithmetic averaging, and finally all aligned, converted, and expanded records form the original operation data set.

[0095] The benefit of the formula is that by comparing the actual energy consumption with the simulated predicted energy consumption and using square and square root operations, the resource utilization efficiency measurement takes into account the degree of match between actual consumption and predicted values ​​while maintaining a quantifiable range of the result value.

[0096] The acquisition steps are as follows: Use the original running data set obtained above to The actual energy consumption in a fixed test cycle is counted and accumulated period by period, and the recorded energy consumption value is in kWh. The actual measured data formed by multiple continuous measurements by the electric energy meter is used as the basis, and the energy consumption is calculated. ;

[0097] The acquisition steps are as follows: query the device through the previously obtained device prediction model and simulation prediction data The predicted energy consumption record within the same test cycle is read out in kWh and confirmed by comparing it with the predicted data file. ;

[0098] Calculation process:

[0099] Substituting the above parameters into the formula:

[0100] ;

[0101] First calculate the numerator:

[0102] ;

[0103] Continue substituting to find the square root:

[0104] ;

[0105] The result of the square root is:

[0106] ;

[0107] This result shows that when When the value is 0.8, the actual consumption of the equipment (0.8kWh) is slightly lower than the predicted value (1.0kWh), which means that the resource utilization efficiency of the equipment in this cycle is lower than that of the fully matched state (corresponding to ),like It means that the actual consumption of the equipment is higher than the predicted value. This means that the actual consumption of the equipment is lower than the predicted value.

[0108] According to the resource utilization efficiency obtained above, the resource utilization efficiency is used as the search keyword to query the corresponding parameter adjustment record in the device parameter configuration file, and the numerical range of each device adjustable parameter and the energy consumption change data after past adjustment are read from the file. Based on the adjustable range of each device parameter, the specific parameter to be adjusted is identified from the resource utilization efficiency value, and the current value of each parameter to be adjusted is numerically modified. During the modification process, the parameter value is compared with the standard value in the device historical calibration data table, and the parameter with the modified value is recorded back to the device parameter configuration file. The modified parameters are reread one by one to check the parameter writing result. If there is no difference, the corrected parameter values ​​are batch-written to the device control execution table, and instructions are sent to the device control terminal through the parameter update instructions given in the execution table. After each instruction is sent, the device feedback signal is read and compared with the specified feedback parameter value in the parameter change table. If there is a deviation, the modification operation is repeated until the feedback signal is consistent with the standard reference. Finally, the records after the parameter adjustment are archived and integrated into an optimized device scheduling plan.

[0109] The steps for obtaining equipment operation effect data are as follows:

[0110] Implement optimized equipment scheduling schemes, monitor the operating status and energy consumption of each device in real time, and obtain real-time monitoring data sets;

[0111] Based on the real-time monitoring data set, the energy efficiency performance index of each device is calculated using the following formula:

[0112] ;

[0113] in, For equipment The energy efficiency performance index, For equipment The total energy consumed during the monitoring period, is the operating output of the equipment during the same period;

[0114] According to the energy efficiency performance index, the operating efficiency of all equipment is summarized and analyzed, and the equipment performance indicators and energy consumption data are integrated to generate equipment operation effect data.

[0115] Specifically, according to the optimized equipment scheduling scheme obtained above, the operating status parameters of each equipment are read cyclically through the independent metering instrument connected to the equipment, the instantaneous power consumption value of the equipment in each time period is recorded in the data table, the value is transmitted from the metering instrument to the data file at intervals of minutes, the read energy consumption value is matched with the corresponding timestamp, the corresponding record row is found with the equipment number as the retrieval key value, the power consumption value, the operating status flag code and the indoor temperature measurement value and humidity measurement value in the environmental sensor data are added to the same row of data one by one, if there are missing records in the process, the row is marked in the data table and filled by cross-comparing the data at other time points in subsequent queries, the recorded user activity event number is converted into the corresponding activity identification value by consulting the user activity event index file and added to the same row record, all data rows are sorted in chronological order and stored in the specified data file, the recording process is executed cyclically, and finally the energy consumption and operating status data of continuous time periods are integrated into a real-time monitoring data set.

[0116] The benefit of the formula is that, by performing a joint logarithmic and square root operation on the total energy consumed by the equipment during the monitoring period and the operating output during the same period, the energy efficiency performance index can take into account the data distribution of larger and smaller numerical ranges, and make a balanced measurement of energy consumption and output, thus providing a clear and easy-to-compare indicator for subsequent analysis.

[0117] The steps of obtaining the parameters are as follows: screening the energy consumption and output data of the equipment in the real-time monitoring data set obtained previously;

[0118] The acquisition steps are as follows: In the real-time monitoring data set, the equipment is monitored by measuring instruments. The total energy consumption value of the equipment is obtained by adding up the hourly energy consumption values ​​recorded during the monitoring period (for example, 24 hours). ;

[0119] The steps to obtain the data are as follows: Count the actual output of the equipment during the same monitoring period. The output is measured in quantifiable production quantities and is obtained by accumulating hourly records. ;

[0120] Calculation process:

[0121] Substituting the above parameters into the formula:

[0122] ;

[0123] Calculate the addition part within the brackets:

[0124] ;

[0125] Take the logarithm:

[0126] ;

[0127] Square root calculation:

[0128] ;

[0129] Substitution operation:

[0130] ;

[0131] The results show that when the energy efficiency performance index is 0.844, the relationship between the energy consumption and output of the equipment during the monitoring period is slightly below 1. 1 is a reference value for a relatively balanced energy consumption and output. When the result is greater than 1, it means that the equipment obtains a higher output at the same energy consumption. When the result is less than 0.5, it means that the equipment has a lower output at the same energy consumption.

[0132] According to the energy efficiency performance index obtained above, the number of each device is used as the retrieval basis to select the energy efficiency performance index value corresponding to the device, and the energy efficiency performance index of each device is matched with the device performance index record in the same period one by one. The equipment performance indicators including continuous operation time record, dynamic load data during operation and output quality detection value measured by the internal sensor of the equipment are read out and numerically compared. The energy consumption data column calibrated for each device in the data table is used as a reference, and the energy efficiency performance index and the corresponding equipment performance index value are compared between columns. By introducing the execution comparison process, the difference between the energy efficiency performance index and the performance index in each record row is calculated, and the difference value is compared with the empirical reference value extracted from the historical data statistics in advance. The empirical reference value is determined by the median extracted from multiple rounds of historical sample data, and the record rows higher than the empirical reference value are marked in the data table, and then the marked data is compared with the corresponding energy consumption column. Finally, all the matched data rows are read out and recorded in the result file, so as to obtain the equipment operation effect data.

[0133] The steps to obtain the updated equipment prediction model are:

[0134] Receive equipment operation effect data, sort out and classify the differences between equipment operation performance indicators and model prediction results, and obtain equipment operation difference data;

[0135] Based on the equipment operation difference data, analyze the correlation pattern between equipment operation and prediction data, adjust the input parameters and structure of the equipment prediction model, and generate an updated model structure;

[0136] Based on the updated model structure, the model performance is retrained and verified, and the model weights and bias values ​​are optimized in combination with real-time data feedback to obtain an updated device prediction model.

[0137] Specifically, receive the equipment operation performance data, classify the power consumption data, performance indicators and environmental parameters during the equipment operation according to the equipment category, determine the key parameters including the actual operation time, the total energy consumption and output power of the equipment, and calculate the energy efficiency deviation by comparing the energy efficiency target in the model prediction with the actual operation results of the equipment one by one, and record it in the difference table. For the power consumption data that exceeds the performance expectations of the equipment, use the energy efficiency standard reference table to filter out the data that does not meet the safety threshold, where the safety threshold is determined by the average value of the historical energy efficiency records of the equipment with a fluctuation of 10%. For the records of abnormal temperature and power consumption, further compare the change range of environmental parameters to check whether it exceeds the safety range of the equipment design. For the equipment operation data with high-frequency repeated deviations, mark them as important difference items separately, and finally complete the generation of equipment operation difference data.

[0138] Based on the equipment operation difference data, the performance and energy efficiency deviations of each device during operation are analyzed. First, the causes of the differences are checked for key difference items, and the influence of external conditions such as ambient temperature, humidity and equipment load rate is checked. The input parameters used in the equipment prediction model are analyzed item by item verification method to see whether they are comprehensive. The actual operation data of the equipment is associated with the prediction input parameters item by item. Parameters that are not included in the prediction input range but are highly correlated with the equipment operation, such as sudden load increase and short-term power supply fluctuations, are included in the input adjustment range. The output value of each layer of the network in the model is further compared with the actual deviation result. According to the deviation value, the connection method of the prediction model is optimized layer by layer, the number of output prediction nodes and the parameter update strategy are adjusted to complete the update of the model structure.

[0139] According to the updated model structure, historical training data and equipment operation difference data are loaded, and training data sets are generated by loading them one by one. During the model training process, the weight parameters of the model are optimized according to the equipment type. By recording the error values ​​of each batch in training, the batch results with the smallest error are selected as the preliminary updated version of the model. The real-time operation data is loaded for verification. By comparing the difference between the model prediction value and the real-time operation data, the bias parameters of the model are gradually optimized, and the error range is gradually controlled within plus or minus 2% of the model output standard. Finally, the model after training and verification is saved and deployed to the equipment operation prediction platform to form an updated equipment prediction model.

[0140] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent scheduling method for construction equipment based on AI algorithm, characterized in that: The following steps are involved: Collect equipment energy consumption records, user activity data and meteorological data, perform integrated calculations, and generate comprehensive equipment energy consumption data; use the comprehensive equipment energy consumption data to train a model using a convolutional neural network to obtain an equipment prediction model; Based on the equipment prediction model, calculate the operation schedule of each equipment in the current environment and generate a preliminary equipment scheduling plan; Simulate and implement the preliminary equipment scheduling plan, collect operation data, and perform real-time analysis on the operation data, adjust equipment operation parameters, and obtain an optimized equipment scheduling plan; Implementing the optimized equipment scheduling plan, monitoring equipment operation and energy consumption in real time, and generating equipment operation effect data; analyzing the equipment operation effect data, comparing it with the output of the equipment prediction model, adjusting model parameters, and obtaining an updated equipment prediction model; Based on the updated equipment prediction model, regularly recalculate equipment operation scheduling; The steps for obtaining the preliminary equipment scheduling plan are: Receiving predicted operation data of each device from the device prediction model, analyzing current environmental variables, including temperature, humidity, and device usage frequency, to obtain environmental and prediction data sets; Based on the environment and the predicted data set, the operation scheduling priority of each device is calculated using the following formula: ； in, For equipment The predicted energy consumption, For equipment The efficiency coefficient, For equipment The efficiency reduction coefficient is For equipment Adjustment factor to match environmental conditions, Schedule priority for operations; According to the operation scheduling priority, the equipment operation order is arranged in order of priority, and the interaction impact and energy consumption of the equipment are comprehensively considered to generate a preliminary equipment scheduling plan.

2. The intelligent scheduling method for construction equipment based on AI algorithm according to claim 1 is characterized in that: The steps for obtaining the comprehensive data of equipment energy consumption are as follows: Collect energy consumption records of building equipment, user activity data and meteorological data, and perform preprocessing operations on the three types of data, including data format unification and outlier removal, to obtain the preprocessed data set; Based on the preprocessed data set, the energy consumption records, user activity data and meteorological data are correlated and analyzed, and the correlation analysis data is obtained by combining the equipment usage and meteorological conditions; Based on the correlation analysis data, the data is focused on the factors that affect energy consumption through cluster analysis to generate comprehensive equipment energy consumption data.

3. The intelligent scheduling method for construction equipment based on AI algorithm according to claim 1 is characterized in that: The steps for obtaining the equipment prediction model are: Receiving the comprehensive energy consumption data of the equipment, selecting a convolutional neural network, setting the number of network layers, the number of neurons and the activation function, and obtaining a preliminary network model structure; Based on the preliminary network model structure, the comprehensive data of device energy consumption is loaded, a model training process is performed, and a trained network model is obtained by iteratively optimizing network weights and biases; Verification and testing are performed based on the trained network model, and the prediction ability and accuracy of the model are verified using the test set data to obtain the equipment prediction model.

4. The intelligent scheduling method for construction equipment based on AI algorithm according to claim 1 is characterized in that: The steps for obtaining the optimized equipment scheduling plan are: Implementing the preliminary equipment scheduling plan and performing simulation operation, collecting operation data of equipment performance and environmental response, and obtaining an original operation data set; Calculate the resource utilization efficiency of the device based on the original operation data set; According to the resource utilization efficiency, the equipment operating parameters are adjusted to optimize the overall energy efficiency, thereby obtaining an optimized equipment scheduling plan.

5. The intelligent scheduling method for construction equipment based on AI algorithm according to claim 1 is characterized in that: The steps for obtaining the equipment operation effect data are as follows: Implement the optimized equipment scheduling scheme, monitor the operating status and energy consumption of each equipment in real time, and obtain a real-time monitoring data set; Based on the real-time monitoring data set, the energy efficiency performance index of each device is calculated using the following formula: ； in, For equipment The energy efficiency performance index, For equipment The total energy consumed during the monitoring period, is the operating output of the equipment during the same period; According to the energy efficiency performance index, the operating efficiency of all equipment is summarized and analyzed, and the equipment performance indicators and energy consumption data are integrated to generate equipment operation effect data.

6. The intelligent scheduling method for construction equipment based on AI algorithm according to claim 1 is characterized in that: The steps for obtaining the updated equipment prediction model are: Receiving the equipment operation effect data, sorting and classifying the difference items between the equipment operation performance index and the model prediction result, and obtaining the equipment operation difference data; Based on the equipment operation difference data, the correlation pattern between the equipment operation and the prediction data is analyzed, the input parameters and structure of the equipment prediction model are adjusted, and an updated model structure is generated.

7. The intelligent scheduling method for construction equipment based on AI algorithm according to claim 6 is characterized in that: The step of acquiring the updated equipment prediction model also includes: retraining and verifying the model performance according to the updated model structure, optimizing the model weights and bias values ​​in combination with real-time data feedback, and obtaining the updated equipment prediction model.

Images