Intelligent park energy management system and energy efficiency optimization method

Abstract

The invention discloses an intelligent park energy management system and an energy efficiency optimization method, and the system comprises a data collection module which is disposed in a multi-mode sensor network in a park, and collects the power, gas, photovoltaic, and environment temperature and humidity data; the edge calculation module is connected with the data acquisition module and is used for locally preprocessing the original data; and the cloud platform optimization module is used for receiving the output data of the edge calculation module, executing a multi-objective optimization algorithm based on deep reinforcement learning and generating a dynamic scheduling instruction. During implementation, the method at least has the following effects: 1, the energy efficiency is remarkably improved, and the comprehensive energy efficiency is improved by 18%-25% compared with that of a traditional system through a multi-objective optimization algorithm; 2, the response speed is optimized, the edge-cloud collaborative architecture enables the delay of a control instruction to be less than or equal to 50ms, and the high-dynamic scene requirement is met; 3, compatibility and expansibility: multi-protocol equipment access is supported, and the system can be expanded to novel energy sources such as wind power, energy storage and the like;

Description

Technical Field

[0001] This application relates to the field of smart park energy management technology, and specifically to a smart park energy management system and energy efficiency optimization method. Background Art

[0002] During the operation of the park, since there are many enterprises in the park and many electrical equipment are used, it is necessary to collect the energy consumption of each equipment in order to guide electricity consumption and optimize energy use.

[0003] In the existing technology, most of the time, a single sensor is used to collect data, and different energy-consuming devices collect different types of information. For example, for air conditioners, the collected information is power consumption, and for gas, the collected information is flow rate. However, the technical solutions in the existing technology make it difficult to integrate different data and conduct comprehensive energy consumption analysis. Summary of the Invention

[0004] To this end, this application provides an intelligent campus energy management system and energy efficiency optimization method to solve the problems existing in existing technologies that are difficult to integrate different data and difficult to conduct comprehensive energy consumption analysis.

[0005] In order to achieve the above objectives, this application provides the following technical solutions:

[0006] In a first aspect, a smart park energy management system includes:

[0007] Data acquisition module: A multimodal sensor network deployed within the park collects data on electricity, gas, photovoltaics, and ambient temperature and humidity;

[0008] Edge computing module: connected to the data acquisition module, used for local pre-processing of raw data;

[0009] Cloud platform optimization module: receives the output data of the edge computing module, executes the multi-objective optimization algorithm based on deep reinforcement learning, and generates dynamic scheduling instructions;

[0010] Execution control module: adjusts the operating parameters of energy equipment according to scheduling instructions.

[0011] Preferably, the data acquisition module includes:

[0012] Power monitoring unit: uses high-precision smart meters to collect voltage, current, and power factor at 1-second intervals;

[0013] Environmental sensing unit: integrated temperature, humidity, and light intensity sensors, with data fusion error ≤ 2%;

[0014] Equipment status unit: monitors the health status of key equipment through vibration sensors and infrared thermal imagers.

[0015] Preferably, the edge computing module includes:

[0016] Local data lake: stores historical data for the last 24 hours and supports hybrid SQL and NoSQL queries;

[0017] Lightweight AI model: LSTM model trained with the TinyML framework for short-term load forecasting;

[0018] Anomaly Detection Engine: Uses the Isolation Forest algorithm to identify device anomalies in real time.

[0019] Preferably, the optimization algorithm of the cloud platform optimization module includes the following steps:

[0020] State space construction: Input variables include real-time electricity prices, equipment load rates, weather forecasts, and carbon emission constraints;

[0021] Reward function design: Comprehensively optimize energy costs, equipment life, and user comfort for multi-objective optimization;

[0022] Policy update mechanism: Update policy network parameters based on the latest data every 15 minutes.

[0023] Preferably, the execution control module includes:

[0024] Adaptive PID controller: dynamically adjusts PID parameters according to scheduling instructions, with a control accuracy of ±1%;

[0025] Device linkage interface: supports Modbus and BACnet protocols, and is compatible with third-party device access.

[0026] A method for optimizing energy efficiency of a smart park energy management system comprises the following steps:

[0027] Step S1: Acquire real-time energy data and environmental parameters through the data acquisition module;

[0028] Step S2: The edge computing module performs data cleaning and short-term load forecasting;

[0029] Step S3: The cloud platform optimization module generates a dynamic scheduling strategy based on deep reinforcement learning;

[0030] Step S4: The execution control module drives the device to execute the optimal strategy and feeds back actual energy consumption data.

[0031] Preferably, in step S3, the deep reinforcement learning algorithm adopts a PPO framework, the policy network is a 3-layer fully connected neural network, and the value function network and the policy network share underlying parameters.

[0032] Preferably, the step S4 further includes:

[0033] Dynamic weight adjustment: adjust the reward function weight in real time based on the priority set by the user;

[0034] Strategy rollback mechanism: If the actual energy consumption deviates from the predicted value by more than 10%, it will automatically switch to the historical optimal strategy.

[0035] Preferably, the method further comprises the step of generating an energy efficiency report:

[0036] Generate multi-dimensional energy efficiency reports on a daily / weekly / monthly basis, including carbon emission statistics, equipment health scores, and energy saving potential analysis;

[0037] Report visualization: Display dynamic heat maps, trend curves and optimization suggestions through the web interface.

[0038] Compared with the prior art, this application has at least the following beneficial effects:

[0039] When implemented, the present invention has at least the following effects: 1. Significantly improved energy efficiency: through a multi-objective optimization algorithm, the comprehensive energy efficiency is improved by 18% to 25% compared with traditional systems; 2. Optimized response speed: the edge-cloud collaborative architecture makes the control instruction delay ≤50ms, meeting the needs of high-dynamic scenarios; 3. Compatibility and scalability: supports multi-protocol device access and can be expanded to new energy sources such as wind power and energy storage; 4. Reduced maintenance costs: anomaly detection accuracy is ≥95%, and equipment fault troubleshooting time is shortened by 70%; 5. User-friendly interaction: visual reports and adaptive policy adjustments improve management convenience. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To more intuitively illustrate the prior art and the present application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be considered as limiting conditions for implementing the present application; for example, based on the technical concepts disclosed in this application and the exemplary drawings, those skilled in the art are capable of easily making routine adjustments or further optimizations to the addition / reduction / attribution division, specific shapes, positional relationships, connection methods, dimensional ratios, etc. of certain units (components).

[0041] Figure 1 This is a module diagram of the intelligent campus energy management system provided in Example 1 of this application. DETAILED DESCRIPTION

[0042] The present application will be further described below in detail through specific embodiments in conjunction with the accompanying drawings.

[0043] like Figure 1 As shown, a smart park energy management system includes:

[0044] Data acquisition module: A multimodal sensor network deployed within the park collects data on electricity, gas, photovoltaics, and ambient temperature and humidity;

[0045] Edge computing module: connected to the data acquisition module, used to perform local preprocessing on the raw data. Preprocessing operations can include filtering, normalization, anomaly detection, etc.

[0046] Cloud platform optimization module: receives the output data of the edge computing module, executes the multi-objective optimization algorithm based on deep reinforcement learning, and generates dynamic scheduling instructions;

[0047] Execution control module: adjusts the operating parameters of energy equipment (such as air conditioning, lighting, and energy storage batteries) according to scheduling instructions.

[0048] Through the edge-cloud collaborative architecture, data transmission bandwidth occupancy is reduced by more than 30%, and real-time response speed is improved to millisecond level.

[0049] The data acquisition module includes:

[0050] Power monitoring unit: uses high-precision smart meters (the error of naive smart meters is ≤0.5%) to collect voltage, current, and power factor at 1-second intervals;

[0051] Environmental sensing unit: integrated temperature, humidity, and light intensity sensors, with data fusion error ≤ 2%;

[0052] Equipment status unit: monitors the health status of key equipment (such as transformers, compressors, etc.) through vibration sensors and infrared thermal imagers.

[0053] Realize high-precision synchronous collection of multi-source heterogeneous data, and the monitoring data integrity rate is ≥99%.

[0054] The edge computing module includes:

[0055] Local data lake: stores historical data for the last 24 hours and supports hybrid SQL and NoSQL queries;

[0056] Lightweight AI model: An LSTM model trained with the TinyML framework for short-term load forecasting, ensuring a forecast error of ≤5%;

[0057] Anomaly Detection Engine: Uses the Isolation Forest algorithm to identify device anomalies in real time and ensures an accuracy rate of ≥95%.

[0058] Through the above technical solution, local computing resource usage is reduced by 40%, and the abnormal event detection delay is ≤200ms.

[0059] The optimization algorithm of the cloud platform optimization module includes the following steps:

[0060] State space construction: Input variables include real-time electricity prices, equipment load rates, weather forecasts, and carbon emission constraints;

[0061] Reward function design: Comprehensive energy cost, equipment life, and user comfort (for example, user comfort can be evaluated by introducing the PMV index) for multi-objective optimization;

[0062] Policy update mechanism: The policy network parameters are updated every 15 minutes based on the latest data. By optimizing the cloud platform module, this solution can improve overall energy efficiency by 18% to 25% compared to traditional rule-based scheduling.

[0063] The execution control module includes:

[0064] Adaptive PID controller: Dynamically adjusts PID parameters (Kp, Ki, Kd) according to scheduling instructions, with a control accuracy of ±1%;

[0065] Device linkage interface: supports Modbus and BACnet protocols, is compatible with third-party device access, has a device control response time of ≤50ms, and compatibility covers more than 90% of mainstream brands, making this system more widely applicable and easier to promote and use during implementation.

[0066] A method for optimizing energy efficiency of a smart park energy management system comprises the following steps:

[0067] Step S1: Acquire real-time energy data and environmental parameters through the data acquisition module;

[0068] Step S2: The edge computing module performs data cleaning and short-term load forecasting;

[0069] Step S3: The cloud platform optimization module generates a dynamic scheduling strategy based on deep reinforcement learning;

[0070] Step S4: The execution control module drives the device to execute the optimal strategy and feeds back actual energy consumption data.

[0071] When the above technical solution is implemented, full-process automated closed-loop management can be achieved, and the need for manual intervention can be reduced by 80%, thereby optimizing energy use.

[0072] In step S3, the deep reinforcement learning algorithm adopts the PPO (proximal policy optimization) framework, the policy network is a 3-layer fully connected neural network (256-128-64 nodes), and the value function network and the policy network share the underlying parameters, so that when this method is implemented, the algorithm convergence speed is accelerated and the local optimal trap is avoided.

[0073] The step S4 further includes:

[0074] Dynamic weight adjustment: real-time adjustment of reward function weights based on user-defined priorities (economy > comfort > environmental protection);

[0075] Strategy rollback mechanism: If the actual energy consumption deviates from the predicted value by more than 10%, it will automatically switch to the historical optimal strategy.

[0076] The system robustness is enhanced, and energy consumption fluctuations in extreme scenarios are reduced by 50%.

[0077] The method further comprises the step of generating an energy efficiency report:

[0078] Generate multi-dimensional energy efficiency reports on a daily / weekly / monthly basis, including carbon emission statistics, equipment health scores, and energy saving potential analysis;

[0079] Report visualization: Display dynamic heat maps, trend curves and optimization suggestions through the web interface.

[0080] Managers’ decision-making efficiency increased by 60%, and the implementation cycle of energy-saving measures was shortened by 30%.

[0081] The technical features of the above embodiments can be combined arbitrarily (as long as there is no contradiction in the combination of these technical features). In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described; these embodiments that are not explicitly written should also be considered to be within the scope of this specification.

Claims

1. A smart park energy management system, characterized by: include: Data acquisition module: A multimodal sensor network deployed within the park collects data on electricity, gas, photovoltaics, and ambient temperature and humidity; Edge computing module: connected to the data acquisition module, used for local pre-processing of raw data; Cloud platform optimization module: receives the output data of the edge computing module, executes the multi-objective optimization algorithm based on deep reinforcement learning, and generates dynamic scheduling instructions; Execution control module: adjusts the operating parameters of energy equipment according to dispatch instructions, The data acquisition module includes: Power monitoring unit: uses high-precision smart meters to collect voltage, current, and power factor at 1-second intervals; Environmental sensing unit: integrated temperature, humidity, and light intensity sensors, with data fusion error ≤ 2%; Equipment status unit: monitors the health status of key equipment through vibration sensors and infrared thermal imagers. The edge computing module includes: Local data lake: stores historical data for the last 24 hours and supports hybrid SQL and NoSQL queries; Lightweight AI model: LSTM model trained with the TinyML framework for short-term load forecasting; Anomaly Detection Engine: Uses the Isolation Forest algorithm to identify device anomalies in real time.

2. The intelligent park energy management system according to claim 1, characterized in that: The optimization algorithm of the cloud platform optimization module includes the following steps: State space construction: Input variables include real-time electricity prices, equipment load rates, weather forecasts, and carbon emission constraints; Reward function design: Comprehensively optimize energy costs, equipment life, and user comfort for multi-objective optimization; Policy update mechanism: Update policy network parameters based on the latest data every 15 minutes.

3. The intelligent park energy management system according to claim 1, characterized in that: The execution control module includes: Adaptive PID controller: dynamically adjusts PID parameters according to scheduling instructions, with a control accuracy of ±1%; Device linkage interface: supports Modbus and BACnet protocols, and is compatible with third-party device access.

4. An energy efficiency optimization method for a smart park energy management system based on any one of claims 1-3, characterized in that: The following steps are involved: Step S1: Acquire real-time energy data and environmental parameters through the data acquisition module; Step S2: The edge computing module performs data cleaning and short-term load forecasting; Step S3: The cloud platform optimization module generates a dynamic scheduling strategy based on deep reinforcement learning; Step S4: The execution control module drives the device to execute the optimal strategy and feeds back actual energy consumption data.

5. The energy efficiency optimization method of a smart park energy management system according to claim 4 is characterized in that: In step S3, the deep reinforcement learning algorithm adopts the PPO framework, the policy network is a 3-layer fully connected neural network, and the value function network and the policy network share underlying parameters.

6. The energy efficiency optimization method of a smart park energy management system according to claim 4, characterized in that: The step S4 further includes: Dynamic weight adjustment: adjust the reward function weight in real time based on the priority set by the user; Strategy rollback mechanism: If the actual energy consumption deviates from the predicted value by more than 10%, it will automatically switch to the historical optimal strategy.

7. The energy efficiency optimization method of a smart park energy management system according to claim 4, characterized in that: The method further comprises the step of generating an energy efficiency report: Generate multi-dimensional energy efficiency reports on a daily / weekly / monthly basis, including carbon emission statistics, equipment health scores, and energy saving potential analysis; Report visualization: Display dynamic heat maps, trend curves and optimization suggestions through the web interface.

Images