Intelligent park electric power consumption prediction method and system based on artificial intelligence

Abstract

The invention discloses a smart park electric power consumption prediction method and system based on artificial intelligence, and relates to the field of electric power prediction.The method comprises the steps that multi-source heterogeneous energy consumption data of a smart park is collected and preprocessed, and a migration sample library is constructed based on the preprocessed multi-source heterogeneous energy consumption data; constructing an energy consumption space-time prediction model fusing the time dimension and the space dimension, training the energy consumption space-time prediction model by using the sample data in the migration sample library, predicting the power consumption of the smart park in the future time period through the trained energy consumption space-time prediction model, and obtaining a power consumption prediction result; and an early warning and regulation and control mechanism is constructed based on the electric power consumption prediction result of the smart park, and electric power resource management of the smart park is realized through the early warning and regulation and control mechanism. According to the method, the energy consumption change trend of park power in different time and space dimensions can be accurately predicted, and potential energy consumption abnormities can be found in advance.

Description

Technical Field

[0001] The present invention relates to the technical field of power forecasting, and in particular to a method and system for forecasting power consumption in a smart park based on artificial intelligence. Background Art

[0002] With the rapid advancement of information technology, the widespread application of technologies such as the Internet of Things, big data, artificial intelligence (AI) and cloud computing is driving the transformation of traditional parks towards smart parks. By integrating advanced communication and information technology, smart parks have optimized resource allocation, improved operational efficiency, and played a positive role in promoting the sustainable development of cities. However, with the continuous expansion of park scale and the increasing diversification of electricity demand, the management and prediction of power consumption has become a major challenge that needs to be urgently addressed in the construction of smart parks. In particular, the spatiotemporal variation characteristics of power consumption make it difficult for traditional power management methods to effectively respond to the increasingly complex energy-saving needs of park electricity consumption.

[0003] Traditional power consumption management methods are mostly based on static statistical models or time series analysis. These methods focus on the processing of historical power data, but ignore the spatial differences and temporal fluctuations in power consumption within the park. In particular, with the increasing diversification of buildings, facilities and equipment within the park, the spatial distribution and temporal changes in power demand show more complex spatiotemporal dynamic characteristics. Traditional methods often find it difficult to achieve accurate predictions when dealing with such spatiotemporal changes.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides an artificial intelligence-based smart park power consumption prediction method and system, which has the advantage of combining the spatial differences and temporal fluctuations of power consumption to predict the power consumption of the park, thereby solving the problem of spatial differences and temporal fluctuations of power consumption within the park.

[0006] In order to achieve the advantages of combining the spatial differences in power consumption and the temporal fluctuation characteristics of power consumption to predict the power consumption of the park, the specific technical solutions adopted by the present invention are as follows: According to one aspect of the present invention, a method for predicting power consumption of a smart park based on artificial intelligence is provided, the method comprising: Collect multi-source heterogeneous energy consumption data of the smart park, pre-process the multi-source heterogeneous energy consumption data, and build a migration sample library based on the pre-processed multi-source heterogeneous energy consumption data; Construct an energy consumption spatio-temporal prediction model that integrates the time dimension and the space dimension, and use the sample data in the transfer sample library to train the energy consumption spatio-temporal prediction model. Predict the electricity consumption of the smart park in the future time period through the trained energy consumption spatio-temporal prediction model to obtain the electricity consumption prediction result; Construct an early warning and regulation mechanism based on the electricity consumption prediction result of the smart park, and realize the management of the electricity resources of the smart park through the early warning and regulation mechanism.

[0007] According to another aspect of the present invention, there is also provided a smart park electricity consumption prediction system based on artificial intelligence, and the system includes: A sample library construction module, configured to collect multi-source heterogeneous energy consumption data of the smart park, preprocess the multi-source heterogeneous energy consumption data, and construct a transfer sample library based on the preprocessed multi-source heterogeneous energy consumption data; An electricity consumption prediction module, configured to construct an energy consumption spatio-temporal prediction model that integrates the time dimension and the space dimension, and use the sample data in the transfer sample library to train the energy consumption spatio-temporal prediction model. Predict the electricity consumption of the smart park in the future time period through the trained energy consumption spatio-temporal prediction model to obtain the electricity consumption prediction result; An electricity resource management module, configured to construct an early warning and regulation mechanism based on the electricity consumption prediction result of the smart park, and realize the management of the electricity resources of the smart park through the early warning and regulation mechanism.

[0008] Compared with the prior art, the present invention provides a smart park electricity consumption prediction method and system based on artificial intelligence, and has the following beneficial effects: The present invention can accurately predict the energy consumption change trend of the park electricity in different time and space dimensions, discover potential energy consumption anomalies in advance and make effective adjustments, thereby improving the accuracy of energy consumption management, enhancing the stability and reliability of the system, and dynamically optimizing the operation of equipment based on real-time data and historical records to ensure that in a changing environment and demand, the energy consumption always maintains the optimal state, thereby effectively improving the energy utilization efficiency, reducing carbon emissions, and providing a solid technical guarantee for achieving the goal of green energy development. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0010] Figure 1 is a flowchart of a smart park electricity consumption prediction method based on artificial intelligence according to an embodiment of the present invention; Figure 2It is a schematic block diagram of an intelligent park power consumption prediction system based on artificial intelligence according to an embodiment of the present invention.

[0011] In the figure: 1. Sample library construction module; 2. Power consumption prediction module; 3. Power resource management module. Detailed implementation manners

[0012] To further illustrate the embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be combined with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, those of ordinary skill in the art should be able to understand other possible implementation manners and the advantages of the present invention.

[0013] According to an embodiment of the present invention, there is provided an intelligent park power consumption prediction method and system based on artificial intelligence.

[0014] Now, the present invention will be further described in conjunction with the accompanying drawings and specific implementation manners. As Figure 1 shown, an intelligent park power consumption prediction method based on artificial intelligence according to an embodiment of the present invention, the method includes: S1. Collect multi-source heterogeneous energy consumption data of the intelligent park, preprocess the multi-source heterogeneous energy consumption data, and construct a migration sample library based on the preprocessed multi-source heterogeneous energy consumption data.

[0015] Among them, collecting multi-source heterogeneous energy consumption data of the intelligent park, preprocessing the multi-source heterogeneous energy consumption data, and constructing a migration sample library based on the preprocessed multi-source heterogeneous energy consumption data includes: Use Internet of Things sensing devices to collect multi-source heterogeneous energy consumption data of the intelligent park, and the multi-source heterogeneous energy consumption data includes power equipment operation data, environmental variable data, and personnel activity data; It should be noted that the present invention fully considers the characteristics of multi-source data in the intelligent park and comprehensively collects data in multiple dimensions such as IoT device operation data, environmental information, and historical load records. Among them, the device operation data covers basic parameters such as power consumption, voltage, and current, truly reflecting the power consumption patterns of various devices inside the park; the environmental information covers external environmental factors such as temperature and humidity, which helps to deeply analyze the potential impact of environmental changes on power demand; the historical load records are used to reveal the periodic and trend characteristics of energy consumption. In addition, the present invention also incorporates personnel activity data to reflect the spatio-temporal correlation between the crowd density in the park and power consumption. By integrating and uniformly modeling these multi-dimensional data, a high-quality feature matrix is provided for the prediction model, thereby significantly enhancing the adaptability and prediction accuracy of the model.

[0016] Use the outlier detection algorithm to eliminate the abnormal data caused by environmental interference factors in the multi-source heterogeneous energy consumption data, and fill the missing data in the multi-source heterogeneous energy consumption data through the linear interpolation algorithm; Perform normalization and noise reduction processing on the multi-source heterogeneous energy consumption data in sequence to eliminate the data dimension difference and high-frequency noise interference, and obtain the preprocessed multi-source heterogeneous energy consumption data; Reconstruct the preprocessed multi-source heterogeneous energy consumption data to form a spatio-temporal data matrix reflecting the power consumption law of the smart park, and calculate the power sample consumption of the smart park according to the spatio-temporal data matrix, and integrate the calculation results to generate core feature sample data; It should be noted that for the collected Internet of Things (IoT) multi-source heterogeneous data, the present invention implements a series of cleaning, transformation, and reconstruction processes. Specifically, outlier elimination and missing value interpolation methods are used to address data quality issues, and normalization processing and noise reduction techniques are applied to eliminate data dimension differences and high-frequency noise interference. Finally, through the reconstruction processing in the time dimension and space dimension, a high-dimensional spatio-temporal data matrix is successfully constructed, ensuring that the model can accurately capture the dynamic change law of power consumption.

[0017] Construct a migration sample library based on the core feature sample data, and trigger cross-level sample migration in the migration sample library according to the predefined power consumption prediction requirements.

[0018] It should be noted that the energy consumption spatio-temporal prediction model of the present invention aims to predict the power consumption information of each area (space dimension) and different time periods (time dimension) in the smart park. The specific steps for constructing the sample library required for model training are as follows: 1. The primary task of constructing the sample library is to conduct comprehensive and multi-dimensional power consumption data collection. By deploying smart meters, environmental sensors, and pedestrian flow monitoring devices in the park, obtain the operating parameters such as the power consumption, voltage, and current of the devices, and at the same time collect environmental variable data such as temperature, humidity, and light intensity. In addition, by connecting with the weather forecast system and the park historical data system, obtain weather change and historical load trend information. All the collected data are uploaded and integrated in real time through the Internet of Things platform, providing multi-source heterogeneous data support for subsequent analysis.

[0019] 2. To ensure data quality, before constructing the sample library, comprehensive preprocessing of the original data is required. First, outlier detection techniques are used to eliminate abnormal data caused by equipment failures or environmental interferences, and linear interpolation is applied to fill in the missing data. Second, the data is normalized to ensure the unity of data with different dimensions; for noise data, methods such as low-pass filtering or wavelet denoising are used for cleaning. In addition, according to the time and space dimensions, the data is segmented and labeled (such as divided into 15-minute, 1-hour, 1-day, 1-week time segments, and divided into equipment levels and functional areas in terms of space) to form a preprocessed dataset that can reflect spatio-temporal characteristics.

[0020] 3. After completing the data preprocessing, it is necessary to calculate the sample energy consumption of various equipment and regions to generate the core feature sample data for predicting model training. Based on the collected data such as equipment power consumption, voltage, and current, the real-time power and cumulative energy consumption are calculated. At the same time, considering the environmental conditions (such as the impact of temperature on air conditioner energy consumption) and the intensity of personnel activities (such as the increase in equipment load during peak pedestrian flow periods), the energy consumption data is corrected through a multi-variable weighting method to form an energy consumption index that is closer to the actual electricity consumption scenario. The results of the energy consumption calculation will be an important part of the subsequent sample library, providing accurate numerical features for model input.

[0021] 4. According to the requirements of smart campus power energy consumption prediction, a sample library with multiple levels and functions is constructed. In the time dimension, sub-libraries are established separately according to short-term, medium-term, and long-term needs of the sample library to meet the requirements of different prediction tasks; in the space dimension, it covers the electricity consumption data of the entire campus, building areas, and individual equipment (equipment categories). In addition, the sample library adopts a structured storage method, storing energy consumption data, environmental data, and personnel activity data in layers, and realizing efficient retrieval and management through a time series database. The sample library also designs a dynamic update mechanism, regularly updating and supplementing according to the real-time collected data, and optimizing sample annotation in combination with user feedback to improve the timeliness and accuracy of the data.

[0022] 5. After the sample library is constructed, the labeled data in it is used for the training and verification of the artificial intelligence spatio-temporal prediction model. The training set, verification set, and test set are divided by the method of stratified sampling to ensure the representativeness of the data distribution. During the model training process, time series features are extracted based on the improved long short-term memory network (LSTM) to capture the long-term and short-term dependencies in the time series data. In the verification stage, the generalization ability of the model is evaluated through cross-validation, and the prediction accuracy of the model is measured by error analysis indicators (such as mean square error, mean absolute error, etc.). The verification results will further guide the optimization of the sample library and the adjustment of model parameters to ensure that the model can efficiently adapt to the complex scenarios of smart campus power energy consumption.

[0023] Among them, constructing a migration sample library based on the core feature sample data, and triggering cross-level sample migration in the migration sample library according to the predefined power consumption prediction requirements includes: Constructing an initial sample library based on the core feature sample data and classifying and storing the core feature sample data, and dividing the initial sample library into several sub-libraries according to the classified sample categories in the time dimension and the space dimension; Constructing hierarchical migration trees in the sub-libraries respectively, and calculating the migration offset between the current hierarchical migration tree and each of the remaining hierarchical migration trees.

[0024] It should be noted that the hierarchical migration tree is a tree structure used to represent the migration process of data or tasks from one level to another in a multi-level system. In the hierarchical migration tree, each node represents a level in the multi-level system, and each edge represents the data or task transfer relationship between different levels.

[0025] Among them, calculating the migration offset between the current hierarchical migration tree and each of the remaining hierarchical migration trees includes: Extracting a target sample set from the core feature sample data based on the spatio-temporal data matrix, calculating the total score value of each gene sample in the target sample set, and sorting the total score values of each gene sample to obtain a gene sequence; Extracting a non-target class sample set from the core feature sample data, calculating the total score value of each non-gene sample in the non-target class sample set, and sorting the total score values of each non-gene sample to obtain a non-gene sequence; It should be noted that extracting a target sample set from the core feature sample data and calculating the total score of the target sample includes: constructing a spatio-temporal data matrix, where each row of the spatio-temporal data matrix represents a sample and each column represents a feature (such as a certain environmental factor or biomarker); selecting the core features related to the target class samples in the spatio-temporal data matrix; calculating the total score of the target samples based on the core features; normalizing the scores of each target sample so that the scores are comparable; sorting the scores of all target samples and obtaining the ranking of gene samples according to the score level.

[0026] Using the correlation analysis algorithm to calculate the ranking difference between the gene sequence and the non-gene sequence in the gene dimension, and taking the ranking difference as the migration offset between the current hierarchical migration tree and each of the remaining hierarchical migration trees.

[0027] It should be noted that the correlation analysis algorithm is a gene ranking difference analysis algorithm, and the calculation formula for the ranking difference between the gene sequence and the non-gene sequence in the gene dimension is: ; In the formula, WGS represents the ranking difference between the gene sequence and the non-gene sequence in the gene dimension; Represents the total score value of the i th gene sample in the gene sequence; Represents the total score value of the i th non-gene sample in the non-gene sequence; α Represents the attenuation coefficient, which is used to control the weight of highly ranked genes; n Represents the total number of gene samples; Represents the ranking position of the i th gene sample in the gene sequence; Represents the ranking position of the j th gene sample in the gene sequence; e Represents a constant.

[0028] Based on the sample migration offset, the samples covered by the rule feature with the highest confidence are migrated from the current sub-library to the adjacent sub-library. After the migration is completed, the new rule features of the remaining samples are calculated and used as the basis for migration for a new round of iterative migration until the number of remaining samples is empty or less than the set threshold; before the iterative migration of samples, a hierarchical migration tree classifier is constructed, and the priority order of sample migration is determined according to the predefined power consumption prediction requirements. The hierarchical migration tree classifier triggers the sample migration in the sub-library according to the priority order.

[0029] S2. Construct an energy consumption spatio-temporal prediction model that integrates the time dimension and the space dimension, and use the sample data in the migrated sample library to train the energy consumption spatio-temporal prediction model. Predict the power consumption of the smart park in the future time period through the trained energy consumption spatio-temporal prediction model to obtain the power consumption prediction result.

[0030] It should be noted that in order to achieve the goal of high-precision prediction, the present invention constructs a park energy consumption spatio-temporal prediction model that integrates time series analysis and space modeling techniques. This model uses an improved long short-term memory network (LSTM) to extract the time features of equipment power consumption, accurately capture its periodic fluctuations and long-term trends; thereby realizing the spatio-temporal joint prediction of the park's power consumption. This model introduces a multi-factor weight fusion mechanism, comprehensively considering factors such as environmental variables, equipment parameters, and historical load data, to ensure that the model can flexibly adapt to various complex scenarios.

[0031] Among them, constructing an energy consumption spatio-temporal prediction model that integrates the time dimension and the space dimension, and using the sample data in the migrated sample library to train the energy consumption spatio-temporal prediction model, and predicting the power consumption of the smart park in the future time period through the trained energy consumption spatio-temporal prediction model to obtain the power consumption prediction result includes: Construct an energy consumption spatio-temporal prediction model for park power based on an improved long short-term memory network and a superposition theorem layer network; Among them, the improved long short-term memory network is used to extract the time series features of power consumption to capture the periodic change trend of power consumption. The interpretable performance consumption activation function of the superposition theorem layer network is used to reduce the number of parameters of the energy consumption spatio-temporal prediction model.

[0032] It should be noted that the present invention applies a customized intelligent AI model to the energy consumption prediction of a smart park. During the iterative optimization process of the model, an online learning mechanism is introduced into the iterative training of the model, enabling the model to receive real-time data from Internet of Things (IoT) sensors in real time and automatically adjust the weight parameters through training. This self-optimization process ensures the real-time performance and adaptability of the model, enabling it to dynamically adjust according to changes in the park environment, thereby continuously maintaining an efficient energy consumption prediction ability. This dynamic learning mode further improves the sensitivity of the model to changes in the park operation scenario, promoting the continuous optimization and development of the energy consumption prediction technology, which specifically includes: Model structure: The intelligent AI model adopts an improved LSTM architecture, aiming to enhance the spatio-temporal modeling ability of energy consumption prediction in smart parks. By combining the time characteristics of time series data with the spatial features of application scenarios, the model can accurately predict the energy consumption level. The model includes the following innovative structures.

[0033] Improved scalar long short-term memory (S-LSTM): The model innovatively introduces an exponential gating and memory mixing mechanism, enhancing the performance of traditional LSTM networks in processing time series data. S-LSTM not only supports multiple memory units but also realizes the mixing of memory units through recursive connections, enabling the model to effectively extract complex patterns and make accurate predictions when facing tasks that require long-term state tracking.

[0034] In the energy consumption prediction of smart parks, power usage data is typical time series scalar data. The S-LSTM network can accurately capture the time dependence of the data, thereby improving the accuracy of energy consumption prediction. By introducing exponential gating, the model can flexibly adjust the decisions of storage and forgetting to ensure adaptability to time dynamic changes. The introduction of the memory mixing mechanism enables multiple memory units to work together, further enhancing the ability to extract complex spatio-temporal patterns.

[0035] Superposition Theorem Layer Network (Kolmogorov - Arnold Network): Different from the traditional neural network that uses fixed activation functions, the superposition theorem model introduces learnable activation functions at the edges of the network. For the application scenario of energy - consumption multi - source heterogeneous data, this invention designs specialized energy - consumption activation functions, which are usually parameterized by non - linear energy - consumption piece - wise functions, thus providing the model with high flexibility. Each energy - consumption weight parameter is replaced by a univariate function, enabling the model to express complex functions with fewer parameters, which not only improves the flexibility of the model but also enhances its interpretability. When facing complex tasks, this method significantly improves the computational efficiency of the model.

[0036] In this model, the superposition theorem layer significantly improves the performance of energy - consumption prediction in smart campuses by accurately modeling spatial dependence relationships. This layer deeply processes the input data through recursive connections and non - linear transformations, effectively extracting high - dimensional spatial features. Compared with the traditional multi - layer perceptron (MLP), the superposition theorem layer not only excels in reducing the number of model parameters but also effectively reduces the storage requirements of the system, thus optimizing the computational efficiency.

[0037] The AI model of this invention combines the scalar long - short - term memory (S - LSTM) module with the KAN (Kolmogorov–Arnold Network) layer, giving full play to the advantages of both and forming a powerful spatio - temporal prediction system. In the S - LSTM module, the exponential gating and memory mixing mechanism strengthen the modeling ability of long - term and short - term dependencies in multi - source energy - consumption scalar time - series data, enabling the model to accurately predict the dynamic changes in campus energy consumption. The KAN layer, with its new network perspective and spatial modeling ability, effectively captures the complex spatial dependence relationships between campus devices and regions, further improving the model's performance in spatial feature learning. The KAN layer also optimizes the system storage and computational efficiency by reducing the number of parameters. The combination of the three enables this model to not only have a powerful spatio - temporal modeling ability but also provide accurate energy - consumption predictions in real - time in complex and dynamic campus environments.

[0038] Judge whether there is a prediction deviation in the electricity energy - consumption prediction result. If there is no prediction deviation, then use the electricity energy - consumption prediction result output by the energy - consumption spatio - temporal prediction model as the final result. Otherwise, correct the deviation in the electricity energy - consumption prediction result based on the potential superposition effect between sample features to obtain the corrected electricity energy - consumption prediction result.

[0039] Among them, correcting the deviation in the electricity energy - consumption prediction result based on the potential superposition effect between sample features to obtain the corrected electricity energy - consumption prediction result includes: Extract the sample features after classifying the core feature sample data, and the sample features include the operation features of power equipment, environmental variable features, and personnel activity features; Aggregate the operating states of adjacent power equipment based on the operating features of power equipment, and analyze the power equipment jump deviation existing in the start and stop of power equipment during the aggregation of adjacent power equipment operating states.

[0040] It should be noted that for the quantification of jump deviation, an improved Hampel filtering algorithm is used to identify the abnormal residuals in the operating state residuals, and the jump deviation index is defined as: ; In the formula, D k represents the jump deviation index of power equipment k , represents the operating state residual of power equipment k . When the start and stop event of a certain power equipment is detected (judged by the power change rate threshold), the operating state residuals of all adjacent equipment within 5 seconds before and after the event are extracted , represents the measured operating state (such as the instantaneous current value) of power equipment k within the start and stop time window, represents the operating state of power equipment k predicted by the LSTM model under steady state. Δ represents the time window (when the start and stop event of power equipment is detected, 5 seconds are taken before and after the event to form a time window with a total length of 10 seconds). Among them, the model input is the current and voltage sequences during the non-mutation period of the equipment history steady state, and the model output is the predicted steady state operating value at the current moment. When ΔS k > 0, the actual operating state is higher than expected, which may be due to sudden load increase or equipment abnormality. When ΔS k < 0, the actual operating state is lower than expected, which may be due to equipment shutdown or power supply interruption; μ and σ represent the mean and standard deviation of the residuals of adjacent equipment. When it exceeds 3 σ , it is judged as a significant jump; finally, through the backpropagation correction mechanism, the jump deviation is fed back to the equipment control strategy to dynamically adjust the protection setting value or power distribution weight of adjacent equipment.

[0041] Use the improved particle filter algorithm to simulate the superposition effect between the operating features of power equipment and environmental variable features, and capture the mutual adaptation deviation between the operation of power equipment and environmental factors.

[0042] Among them, using the improved particle filter algorithm to simulate the superposition effect between the operating features of power equipment and environmental variable features, and capturing the mutual adaptation deviation between the operation of power equipment and environmental factors includes: Randomly generate a number of particles from the operation characteristics of power equipment and environmental variable characteristics, and assign initial weights. Construct a state transition equation to analyze the current particle state quantity, and predict the particle state quantity at the next time point based on the first-order Markov process; Compare the difference between the particle state at the next time point and the current particle state quantity with a preset threshold, update the particle weights based on the comparison result, and calculate the adaptation relationship between the operation characteristics of power equipment and environmental variable characteristics based on the updated particle weights; Reconstruct the particle boost based on the adaptation relationship, construct a particle state estimator using an improved particle filter algorithm, identify the source strength distribution during the particle reconstruction boost process through the particle state estimator, and identify the mutual adaptation deviation between the operation of power equipment and environmental factors according to the particle source strength distribution.

[0043] It should be noted that the improved particle filter algorithm introduces a comparison between the difference between the particle state and the current state and a preset threshold when updating the particle weights. This mechanism can adjust the particle weights more sensitively, especially in a rapidly changing environment, and can more accurately reflect the changes in the equipment and environmental states.

[0044] Among them, reconstructing the particle boost based on the adaptation relationship, constructing a particle state estimator using an improved particle filter algorithm, identifying the source strength distribution during the particle reconstruction boost process through the particle state estimator, and identifying the mutual adaptation deviation between the operation of power equipment and environmental factors includes: Fuse the particle set during the particle reconstruction boost process, and each particle in the particle set represents the joint state of the operation characteristics of power equipment and environmental variable characteristics; Construct a transfer matrix between the source strength distribution and the particle reconstruction surface based on the joint state, and construct an observation guidance term using the pseudo-inverse of the transfer matrix; Suppress the diffusion range during the particle reconstruction boost process based on the covariance matrix to make the observation guidance term meet the perturbation amplitude condition of the transfer matrix.

[0045] Use the random permutation resampling algorithm to retain the high-weight particles in the particle set that meet the perturbation amplitude condition, and at the same time perform Gaussian perturbation on the low-weight particles in the particle set to obtain an updated particle set; Construct a particle state estimator based on the updated particle set, calculate the weighted average state of the updated particle set through the particle state estimator, and estimate the mutual adaptation deviation between the operation of power equipment and environmental factors according to the weighted average state.

[0046] It should be noted that calculating the weighted average state of the updated particle set through the particle state estimator and estimating the mutual adaptation deviation between the power equipment and environmental factors includes: calculating the weighted average state of the particle set according to the weights of the particle set; estimating the mutual adaptation deviation between the power equipment and environmental factors based on the weighted average state; when judging sudden personnel gathering based on personnel activity characteristics, capturing the superposition effect between the changes in personnel activities and the changes in the operating state of power equipment, and capturing the local load deviation caused by dense personnel, which specifically includes: According to the personnel activity characteristics, judge whether there is a sudden personnel gathering situation, establish a joint model based on the sudden personnel gathering situation, fuse the personnel activity data and the operating state data of the power equipment, and use the joint model to predict the possible local load deviation when the personnel are dense. Dynamically adjust the operating strategy of the power equipment based on this prediction and take corresponding load management measures to cope with the load fluctuation caused by the sudden personnel gathering.

[0047] Fuse the power equipment jump deviation, mutual adaptation deviation and local load deviation to obtain the total deviation of the power consumption prediction result, and use the causal inference algorithm to correct the total deviation in the power consumption prediction result.

[0048] It should be noted that using the causal inference algorithm to correct the total deviation in the power consumption prediction result includes: Step 1: Fuse the power equipment jump deviation, mutual adaptation deviation and local load deviation to obtain the total deviation of the power consumption prediction result; Step 2: Based on the total deviation of the power consumption prediction result, use the causal inference algorithm of kernel density estimation to construct a causal graph model, and obtain the probability distribution of the total deviation through kernel density estimation in the context of causal inference; Step 3: Infer the causal relationship between each pair of variables based on the probability distribution of the total deviation, analyze the strength and direction of the causal relationship between each pair of variables through regression; analyze the causal path between variables, and at the same time, identify the key causal factors through the influence of each variable in the causal path; correct the total deviation in the power consumption prediction result according to the inferred causal relationship and the probability density function of the total deviation.

[0049] S3. Construct an early warning and regulation mechanism based on the power consumption prediction result of the smart park, and realize the power resource management of the smart park through the early warning and regulation mechanism.

[0050] It should be noted that according to the analysis of the prediction results, an intelligent early warning and regulation system is constructed to achieve the efficient management of the park's power resources. When the predicted load touches the preset threshold, the system will automatically trigger a peak warning or an abnormal alarm, and flexibly adjust the operating conditions of the equipment and the load distribution strategy according to the actual situation. Combining the data of load prediction, the system takes measures such as optimizing the power consumption period, allocating green energy, and shutting down unnecessary and inefficiently operating equipment, formulates a regulation plan for energy conservation and consumption reduction, and incorporates the regulation effect into the feedback link to continuously improve the energy consumption management strategy of the park.

[0051] When the predicted load approaches or exceeds the preset threshold, the system will automatically trigger the early warning mechanism and adjust the operating state of the equipment and the load distribution according to the real-time demand. The regulation strategies cover aspects such as peak load balancing, shutting down inefficient equipment, and distributed energy scheduling. At the same time, renewable energy is preferentially used to achieve the goals of energy conservation and emission reduction.

[0052] It should be noted that the core of the artificial intelligence spatio-temporal prediction and intelligent optimization of power energy consumption in the smart park of the present invention lies in the application of artificial intelligence technology in two dimensions of time and space, combined with power demand prediction and energy optimization scheduling, aiming to reduce energy waste and promote the process of green development. Its specific manifestations are as follows: 1. Prediction in the time dimension: The long-term scale prediction realized by the present invention mainly makes a macro estimate of the power demand based on the long-term trends of quarters, years or even multiple years. Such predictions are mostly used for the judgment of the overall energy demand of the park and provide a basis for the annual or quarterly energy planning and budget preparation. The time scale prediction realized by the present invention focuses on the power consumption changes within a week or a month in the park, which helps the park make more accurate short-term energy scheduling decisions. It pays attention to periodic changes, such as the differences between weekdays and weekends, and the changes in activity density at the beginning and end of the month.

[0053] 2. Prediction in the space dimension: The park-wide power demand prediction realized by the present invention mainly focuses on the overall power consumption of the park, which is affected by multiple factors such as weather, seasonal changes, external power grid supply conditions, and overall park activities. Such predictions are mostly used for the annual energy planning and macro-level optimization of the park. The park realized by the present invention can usually be divided into multiple functional areas (such as office areas, commercial areas, industrial areas, parking lots, etc.), and the power demand patterns of each area are significantly different. Based on the differences in the energy consumption laws of spatial regions, more accurate energy consumption information positioning and analysis are carried out.

[0054] 3. Power energy consumption early warning mechanism: The present invention realizes the real-time monitoring and prediction of the power consumption of parks, buildings or equipment, identifies potential abnormal energy consumption phenomena in advance, and issues warning signals in a timely manner. This mechanism aims to optimize energy use, avoid unnecessary waste, and ensure the stable operation of the system.

[0055] The grading is as follows: ① Low-level warning (yellow): When the energy consumption prediction result shows that the energy consumption level of a certain area or equipment is close to the set warning threshold, although it is still within the controllable range, it is still necessary to strengthen monitoring, do a good job in adjustment and optimization, and prevent the trend from deteriorating further.

[0056] ② Medium-level warning (orange): When the energy consumption prediction result shows that the energy consumption of a certain area or equipment has increased significantly, exceeding the normal level and showing a continuous growth trend, if not adjusted in time, it may lead to a further increase in energy consumption, affecting the efficiency and stability of the entire system.

[0057] ③ High-level warning (red): When the energy consumption prediction result shows that the energy consumption of a certain area or equipment has far exceeded the normal level and is close to or exceeds the set target value, it indicates that serious energy waste has occurred. At this time, emergency response measures need to be taken immediately to prevent energy consumption from getting out of control or causing other serious consequences.

[0058] According to another embodiment of the present invention, as Figure 2 shown, there is also provided an intelligent park power consumption prediction system based on artificial intelligence, and the system includes: A sample library construction module 1, which is used to collect multi-source heterogeneous energy consumption data of the intelligent park, preprocess the multi-source heterogeneous energy consumption data, and construct a migration sample library based on the preprocessed multi-source heterogeneous energy consumption data; A power consumption prediction module 2, which is used to construct an energy consumption spatio-temporal prediction model that integrates the time dimension and the space dimension, and use the sample data in the migration sample library to train the energy consumption spatio-temporal prediction model, and predict the power consumption of the intelligent park in the future time period through the trained energy consumption spatio-temporal prediction model to obtain the power consumption prediction result; A power resource management module 3, which is used to construct a warning and regulation mechanism based on the power consumption prediction result of the intelligent park, and realize the power resource management of the intelligent park through the warning and regulation mechanism.

[0059] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting power consumption of a smart park based on artificial intelligence, characterized in that: The method includes: Collect multi-source heterogeneous energy consumption data of the smart park, pre-process the multi-source heterogeneous energy consumption data, and build a migration sample library based on the pre-processed multi-source heterogeneous energy consumption data; Construct a spatiotemporal energy consumption prediction model that integrates the time dimension and the space dimension, and use the sample data in the migration sample library to train the spatiotemporal energy consumption prediction model. The trained spatiotemporal energy consumption prediction model is used to predict the power consumption of the smart park in the future time period to obtain the power consumption prediction result. Based on the power consumption prediction results of the smart park, an early warning and regulation mechanism is established, and the power resource management of the smart park is realized through the early warning and regulation mechanism.

2. According to the artificial intelligence-based smart park power consumption prediction method of claim 1, it is characterized in that: The collecting of multi-source heterogeneous energy consumption data of the smart park, preprocessing of the multi-source heterogeneous energy consumption data, and building of a migration sample library based on the preprocessed multi-source heterogeneous energy consumption data include: Using IoT sensor devices to collect multi-source heterogeneous energy consumption data of smart parks, and the multi-source heterogeneous energy consumption data includes power equipment operation data, environmental variable data and personnel activity data; The outlier detection algorithm is used to remove the abnormal data caused by environmental interference factors in the multi-source heterogeneous energy consumption data, and the linear interpolation algorithm is used to fill the missing data in the multi-source heterogeneous energy consumption data; Perform normalization and noise reduction processing on the multi-source heterogeneous energy consumption data in turn to eliminate data dimension differences and high-frequency noise interference, and obtain pre-processed multi-source heterogeneous energy consumption data; Reconstruct the preprocessed multi-source heterogeneous energy consumption data to form a spatiotemporal data matrix that reflects the power consumption law of the smart park, calculate the power sample energy consumption of the smart park based on the spatiotemporal data matrix, and integrate the calculation results to generate core feature sample data; A migration sample library is constructed based on the core feature sample data, and cross-level sample migration is triggered in the migration sample library according to the predefined power energy consumption forecast requirements.

3. The method for predicting power consumption of a smart park based on artificial intelligence according to claim 2 is characterized in that: The method of constructing a migration sample library based on the core feature sample data and triggering cross-level sample migration in the migration sample library according to predefined power energy consumption prediction requirements includes: Based on the core feature sample data, an initial sample library is constructed and the core feature sample data is classified and stored. According to the classified sample categories, the initial sample library is divided into several sub-libraries according to the time dimension and the space dimension; Build hierarchical migration trees in each sub-database and calculate the migration offset between the current hierarchical migration tree and each of the remaining hierarchical migration trees; Based on the sample migration offset, the samples covered by the rule features with the highest confidence are migrated from the current sub-library to the adjacent sub-library. After the migration is completed, the new rule features of the remaining samples are calculated and used as the basis for migration to perform a new round of iterative migration until the number of remaining samples is empty or less than the set threshold; Before the iterative migration of samples, a hierarchical migration tree classifier is constructed, and the priority order of sample migration is formulated according to the predefined power consumption prediction requirements. The hierarchical migration tree classifier triggers the migration of samples in the sub-library according to the priority order.

4. The method for predicting power consumption of a smart park based on artificial intelligence according to claim 3 is characterized in that: The calculation of the migration offset between the current level migration tree and each of the remaining level migration trees includes: Extract the target sample set from the core feature sample data based on the spatiotemporal data matrix, calculate the total score value of each gene sample in the target sample set, and sort the total score value of each gene sample to obtain the gene sequence; Extract a non-target sample set from the core feature sample data, calculate the total score value of each non-gene sample in the non-target sample set, and sort the total score value of each non-gene sample to obtain a non-gene sequence; The association analysis algorithm is used to calculate the sorting difference between gene sequences and non-gene sequences in the gene dimension, and the sorting difference is used as the migration offset between the current level migration tree and each of the remaining level migration trees.

5. The method for predicting power consumption of a smart park based on artificial intelligence according to claim 1 is characterized in that: The energy consumption spatiotemporal prediction model integrating the time dimension and the space dimension is constructed, and the energy consumption spatiotemporal prediction model is trained using the sample data in the migration sample library. The power consumption of the smart park in the future time period is predicted by the trained energy consumption spatiotemporal prediction model, and the power consumption prediction results obtained include: Based on the improved long short-term memory network and superposition theorem layer network, a spatiotemporal prediction model for the energy consumption of the park power was constructed; The sample data in the migration sample library is divided into training set, validation set and test set by using the stratified sampling algorithm. The power consumption of the smart park in the future time period is predicted through the energy consumption spatiotemporal prediction model after training, and the power consumption prediction result is obtained. Determine whether there is a prediction bias in the power consumption forecast result. If there is no prediction bias, the power consumption forecast result output by the energy consumption spatiotemporal forecast model is used as the final result. Otherwise, the bias in the power consumption forecast result is corrected based on the potential superposition effect between sample features to obtain a corrected power consumption forecast result.

6. The method for predicting power consumption of a smart park based on artificial intelligence according to claim 5 is characterized in that: The correction of the deviation in the power consumption prediction result based on the potential superposition effect between the sample features to obtain the corrected power consumption prediction result includes: Extracting sample features after classification of core feature sample data, wherein the sample features include power equipment operation features, environmental variable features, and personnel activity features; Aggregate the operating status of adjacent power equipment based on the operating characteristics of the power equipment, and analyze the power equipment jump deviation existing in the start and stop of the power equipment in the aggregation of the operating status of adjacent power equipment; The improved particle filter algorithm is used to simulate the superposition effect between the operation characteristics of power equipment and the characteristics of environmental variables, and to capture the mutual adaptation deviation between the operation of power equipment and environmental factors; When judging sudden gatherings of people based on the characteristics of their activities, the superposition effect between changes in people's activities and changes in the operating status of power equipment is used to capture local load deviations caused by dense crowds. The jump deviation of power equipment, mutual adaptation deviation and local load deviation are integrated to obtain the total deviation of power energy consumption forecast results, and the causal reasoning algorithm is used to correct the total deviation in the power energy consumption forecast results.

7. The method for predicting power consumption of a smart park based on artificial intelligence according to claim 6 is characterized in that: The improved particle filter algorithm is used to simulate the superposition effect between the operation characteristics of power equipment and the characteristics of environmental variables, and to capture the mutual adaptation deviation between the operation of power equipment and environmental factors, including: Randomly generate a number of particles from the operating characteristics of power equipment and environmental variables and assign initial weights, construct a state transfer equation to analyze the current particle state quantity, and predict the particle state quantity at the next time point based on the first-order Markov process; Compare the difference between the particle state at the next time point and the current particle state with a preset threshold, update the particle weight based on the comparison result, and calculate the adaptation relationship between the power equipment operation characteristics and the environmental variable characteristics based on the updated particle weight; Particle boosting is reconstructed based on the adaptation relationship, and a particle state estimator is constructed using an improved particle filter algorithm. The source strength distribution in the particle reconstruction boosting process is identified through the particle state estimator, and the mutual adaptation deviation between the operation of power equipment and environmental factors is identified according to the particle source strength distribution.

8. The method for predicting power consumption of a smart park based on artificial intelligence according to claim 7 is characterized in that: The particle boosting is reconstructed based on the adaptation relationship, a particle state estimator is constructed using an improved particle filter algorithm, source intensity distribution in the particle reconstruction boosting process is identified by the particle state estimator, and mutual adaptation deviation between the operation of the power equipment and environmental factors is identified according to the particle source intensity distribution. The method includes: The fusion particles reconstruct the particle set in the boosting process, and each particle in the particle set represents the joint state of the operation characteristics of the power equipment and the characteristics of the environmental variables; Based on the joint state, the transfer matrix between the source intensity distribution and the particle reconstruction surface is constructed, and the observation guidance term is constructed using the pseudo-inverse of the transfer matrix. The diffusion range of the particle reconstruction boost process is suppressed based on the covariance matrix so that the observation guidance term satisfies the perturbation amplitude condition of the transfer matrix; The random permutation resampling algorithm is used to retain the high-weight particles in the particle set that meet the perturbation amplitude conditions, and at the same time, Gaussian perturbations are performed on the low-weight particles in the particle set to obtain the updated particle set; A particle state estimator is constructed based on the updated particle set, a weighted average state of the updated particle set is calculated by the particle state estimator, and the mutual adaptation deviation between the operation of the power equipment and the environmental factors is estimated according to the weighted average state.

9. The method for predicting power consumption of a smart park based on artificial intelligence according to claim 5 is characterized in that: The time series characteristics of power consumption are extracted through the improved long short-term memory network to capture the periodic change trend of power consumption; the interpretable energy consumption activation function of the theorem layer network is superimposed to reduce the number of parameters in the spatiotemporal prediction model of energy consumption.

10. A smart park power consumption prediction system based on artificial intelligence, using the smart park power consumption prediction method based on artificial intelligence as described in any one of claims 1-9, characterized in that: The system includes: The sample library construction module is used to collect multi-source heterogeneous energy consumption data of the smart park, pre-process the multi-source heterogeneous energy consumption data, and build a migration sample library based on the pre-processed multi-source heterogeneous energy consumption data; The power consumption prediction module is used to build a spatiotemporal prediction model of energy consumption that integrates the time dimension and the space dimension, and use the sample data in the migration sample library to train the spatiotemporal prediction model of energy consumption. The trained spatiotemporal prediction model is used to predict the power consumption of the smart park in the future time period to obtain the power consumption prediction result. The power resource management module is used to build an early warning and control mechanism based on the power consumption prediction results of the smart park, and realize the power resource management of the smart park through the early warning and control mechanism.

Images