An artificial intelligence-based energy consumption prediction and optimization system and method

Abstract

The application discloses an energy consumption prediction and optimization system and method based on artificial intelligence, relates to the technical field of energy consumption prediction and optimization, and comprises the following steps: calculating the energy consumption intensity indexes of each device in different time periods to form a time period set of historical energy consumption abnormalities of the device; generating a candidate scheduling strategy, obtaining the device calling sequence and the energy consumption switching time between devices for any strategy, calculating the actual operation energy consumption of the device in the abnormal time period when the scheduling time of the device falls into the historical energy consumption abnormal period set; traversing the device calling time in the strategy to calculate the actual total energy consumption and the number of times of exceeding the predicted energy consumption, calculating the energy efficiency matching value of each candidate strategy, and generating a strategy recommendation sequence; selecting the strategy with the highest energy efficiency matching value from the strategy recommendation sequence as the preferred execution scheme, collecting real-time energy consumption data after execution to calculate the energy consumption deviation rate, and comparing the energy consumption deviation rate with a threshold to determine the effectiveness of the strategy, so that the energy consumption prediction and optimization are realized.

Description

Technical Field

[0001] This invention relates to the field of energy consumption prediction and optimization technology, specifically an energy consumption prediction and optimization system and method based on artificial intelligence. Background Technology

[0002] As commercial facilities continue to expand, enterprises have an increasingly urgent need for refined energy management. As a key cost element in commercial operations, the optimization of equipment energy consumption affects the scientific nature of enterprise operations, green development goals, and business decisions.

[0003] However, traditional energy management methods have gradually revealed the following problems when dealing with complex equipment systems, multi-period energy consumption fluctuations, and the need for intelligent optimization: First, the identification of historical energy consumption anomalies is lagging and lacks systematicity. During the operation of commercial facilities, equipment often experiences energy consumption anomalies due to load fluctuations, equipment aging, or improper scheduling. Traditional management methods rely on manual inspections or post-event statistics, which cannot capture the changes in energy consumption intensity of each device at different times in real time, resulting in long-term energy waste that is difficult to trace. Second, equipment scheduling strategies lack intelligence and dynamic adaptability. Traditional energy consumption scheduling is mostly based on fixed rules and does not combine the historical energy efficiency performance of equipment with real-time needs. In peak-valley electricity pricing scenarios, equipment start-up and shutdown are based solely on time nodes, without considering the actual energy consumption efficiency of different devices at specific times. Third, strategy evaluation and optimization mechanisms lack closed-loop feedback. In traditional energy management, there is a lack of real-time energy efficiency monitoring and dynamic adjustment mechanisms after strategy execution. The actual energy consumption deviation rate exceeds expectations, manual investigation cycles are long and adjustments are lagging, and it is impossible to trigger strategy switching or equipment maintenance in a timely manner, resulting in a gradual decline in the energy efficiency optimization effect. Summary of the Invention

[0004] The purpose of this invention is to provide an energy consumption prediction and optimization system and method based on artificial intelligence to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an energy consumption prediction and optimization method based on artificial intelligence, the method comprising the following steps:

[0006] Acquire historical data of user-specified commercial facilities and equipment, generate a historical data set, calculate the energy consumption intensity index of each device in different time periods, construct an energy consumption intensity change map, mark abnormal time periods when the historical energy consumption intensity is lower than a preset threshold, and form a set of time periods with abnormal historical energy consumption of equipment.

[0007] Based on the user-specified device information and real-time energy consumption requirements, candidate scheduling strategies are generated. For any strategy, the device calling order and the energy consumption switching time between devices are obtained. When the scheduling time of a device falls into its historical energy consumption abnormal segment set, the actual operating energy consumption of the device during the abnormal time period is calculated.

[0008] The actual total energy consumption and the number of times the predicted energy consumption is exceeded are calculated by traversing the device call time in the strategy. The energy efficiency matching value of each candidate strategy is calculated, and the energy efficiency matching value set is generated and sorted from high to low to obtain the strategy recommendation sequence.

[0009] The strategy with the highest energy efficiency matching value is selected from the recommended strategy sequence as the preferred execution plan. Real-time energy consumption data after execution is collected to calculate the energy consumption deviation rate, and the effectiveness of the strategy is determined by comparing it with the threshold.

[0010] The process involves acquiring historical data of user-specified commercial facilities and equipment, generating a historical data set, calculating the energy consumption intensity index of each device over different time periods, constructing an energy consumption intensity change map, marking abnormal time periods where historical energy consumption intensity is lower than a preset threshold, and forming a set of time segments with abnormal historical energy consumption for the equipment. Specific steps include:

[0011] Retrieve historical data of commercial facilities and equipment specified by the user, and generate a historical data set E = {E1, E2, ..., E...} n}; where E1, E2, ..., E n This represents the historical data of devices 1, 2, ..., n, where n represents the total number of commercial facility devices specified by the user. The historical data information of any selected commercial facility device is represented as E. a The historical data information includes: the load rate of device a in the j-th time period, the energy consumption value of device a in the j-th time period, and the operating status of device a in the j-th time period.

[0012] The energy intensity index of device a in the j-th time period is calculated using the following formula: η aj =C aj / (P aj *T j ); where η aj C represents the energy consumption intensity of device i in the j-th time period. aj P represents the energy consumption of device a in the j-th time period. aj T represents the load rate of device i in the j-th time period. j This represents the duration of the j-th time interval;

[0013] The historical data of the target equipment is traversed to obtain the energy consumption intensity index of different equipment in different time periods. A two-dimensional coordinate system is constructed with each time period as the x-axis and the energy consumption intensity index as the y-axis to form an energy consumption intensity variation map of the equipment. Based on the energy consumption intensity variation map, time periods when the historical energy consumption intensity is lower than a preset threshold are marked to form a set Q of time periods with abnormal historical energy consumption of the equipment. a ={Q a1 Q a2 ,...,Q aM '}, where Qa1 Q a2 ,...,Q aM ' indicates that device a has abnormal energy consumption intensity in the 1st, 2nd, ..., M'th time period, and M' represents the total number of time periods in which abnormal energy consumption efficiency was detected.

[0014] Based on user-specified device information and real-time energy consumption requirements, candidate scheduling strategies are generated. For any strategy, the device invocation order and energy consumption switching time between devices are obtained. When the scheduling time of a device falls into its historical energy consumption anomaly segment set, the actual operating energy consumption of the device during the anomaly time segment is calculated. The specific steps include:

[0015] Based on user-specified device information and real-time energy consumption requirements, device scheduling planning is performed, candidate scheduling strategies are generated, and one scheduling strategy is arbitrarily selected and denoted as S. k According to strategy S k The device calling order D(k,i) and the energy switching time between devices T(k,i) are obtained respectively, where T(k,i) represents the time according to strategy S. k The energy adjustment time required to switch from the (i-1)th device to the ith device, D(k,i) represents the time required according to strategy S. k The device called for the i-th time;

[0016] When device D(k,i) is in the time period Located in the historical energy consumption anomaly segment set Q D(k,i) ={Q D(k,i),1 Q D(k,i),2 ,...Q D(k,i),M When the scheduling time falls within a historically inefficient or high-load period of the device, calculate the actual operating energy consumption of device D(k,i) as C. k,i =η D (k,i),j '*P D(k,i),j’ *T j’ ; where η D (k,i),j ' represents the energy consumption intensity of device D(k,i) during the j'-th abnormal time period, P D(k,i),j’ Let T represent the load rate of device D(k,i) during the j'-th abnormal time period, t0 represent the starting time of the policy execution, and T represent the load rate of device D(k,i) during the j'-th abnormal time period. j’ This represents the duration of the j'th abnormal time period.

[0017] The process involves iterating through the device call times in the strategies to calculate the actual total energy consumption and the number of times the predicted energy consumption is exceeded. Then, the energy efficiency matching value for each candidate strategy is calculated, a set of energy efficiency matching values ​​is generated and sorted from high to low, and a strategy recommendation sequence is obtained. Specific steps include:

[0018] Traversal strategy S kCalculate the actual total energy consumption C based on the call time of all devices. k,N And count the number of times K(C) exceeds the predicted energy consumption. k,i >C pred,i ), where C pred,i Indicates device D k,i The predicted energy consumption value, C k,i Indicates device D k,i The actual energy consumption value;

[0019] Computational strategy S k The energy efficiency matching value is defined by the formula shown below: F k =1-K(C k,i >C pred,i ) / N; where K(C k,i >C pred,i The number of times the actual energy consumption of a device exceeds the predicted value in the strategy is represented by , and N represents the total number of devices invoked by the strategy.

[0020] Iterate through all generated candidate policies S1, S2, ..., S R This yields the energy efficiency matching value for each strategy, resulting in a set F = {F1, F2, ..., F...} R}, where R represents the total number of candidate policies generated, F1, F2, ..., F R Let F represent the 1st, 2nd, ..., Rth generated candidate strategies. The obtained set F is sorted from highest to lowest, generating a strategy recommendation sequence F' = {F'1, F'2, ..., F'...}. R}, where F'1>F'2>...>F' R ,F'1,F'2,...,F' R This represents the 1st, 2nd, ..., Rth candidate strategies generated after sorting.

[0021] The strategy with the highest energy efficiency matching value in the recommended strategy sequence is selected as the preferred execution plan. Real-time energy consumption data after execution is collected to calculate the energy consumption deviation rate, which is then compared with a threshold to determine the effectiveness of the strategy. The specific steps include:

[0022] The strategy with the highest energy efficiency matching value is selected from the recommended strategy sequence as the preferred execution plan;

[0023] Real-time energy consumption data after the strategy is executed is collected and compared with the predicted value to calculate the energy consumption deviation rate, defined as follows: Z = |C real -C pred | / C pred *100%; of which, C real C represents the real-time energy consumption data after the strategy is executed. pred represents the predicted energy consumption value, and Z represents the energy consumption deviation rate;

[0024] A threshold Z0 for energy consumption deviation rate is set. When Z≤Z0, the strategy is deemed effective and the current strategy is maintained. When Z>Z0, the strategy is deemed not to have met expectations, and a strategy switching mechanism is triggered. The second-best scoring strategy F'2 is selected from the recommended strategy sequence and updated as the current execution plan. The monitoring and switching process is repeated until either of the following conditions is met: energy consumption deviation rate threshold Z≤Z0, the strategy is executed successfully; all strategies have been tried and have failed to meet the target, an alarm is triggered.

[0025] An AI-based energy consumption prediction and optimization system is applied to the AI-based energy consumption prediction and optimization method described in any one of claims 1-5. The system includes: a data acquisition module, a candidate strategy generation module, a strategy evaluation and ranking module, and a strategy execution and feedback module. The data acquisition module calculates the energy consumption intensity index of each device in different time periods, forming a set of historical energy consumption anomalies for each device. The candidate strategy generation module generates candidate scheduling strategies. For any strategy, it obtains the device calling order and the energy consumption switching time between devices. When the scheduling time of a device falls into its historical energy consumption anomaly set, it calculates the actual operating energy consumption of the device during the anomaly time period. The strategy evaluation and ranking module iterates through the device calling times in the strategies, calculates the actual total energy consumption and the number of times the predicted energy consumption is exceeded, calculates the energy efficiency matching value of each candidate strategy, and generates a strategy recommendation sequence. The strategy execution and feedback module selects the strategy with the highest energy efficiency matching value from the strategy recommendation sequence as the preferred execution plan, collects real-time energy consumption data after execution, calculates the energy consumption deviation rate, and compares it with a threshold to determine the effectiveness of the strategy.

[0026] The data acquisition module includes a data acquisition unit, an energy consumption intensity calculation unit, and an abnormal time period marking unit. The data acquisition unit is used to acquire historical data of user-specified devices and generate a historical data set. The energy consumption intensity calculation unit is used to calculate the energy consumption intensity index of each device in different time periods using a formula. The abnormal time period marking unit is used to construct an energy consumption intensity change graph, compare it with a preset threshold, mark the abnormal time periods when the energy consumption intensity is lower than the threshold, and generate an abnormal time period set.

[0027] The candidate strategy generation module includes a strategy planning unit and an abnormal energy consumption calculation unit. The strategy planning unit is used to generate candidate scheduling strategies based on the device information and real-time energy consumption requirements specified by the user. Each strategy includes the device calling order and switching time. The abnormal energy consumption calculation unit is used to calculate the actual operating energy consumption when the device scheduling time in the strategy falls into its abnormal time segment set.

[0028] The strategy evaluation and ranking module includes an energy consumption statistics unit, an energy efficiency matching calculation unit, and a strategy ranking unit. The energy consumption statistics unit is used to traverse the call time of all devices in the strategy, calculate the actual total energy consumption, and count the number of times the actual energy consumption exceeds the predicted value. The energy efficiency matching calculation unit is used to calculate the energy efficiency matching value of the strategy. The strategy ranking unit is used to sort the set of energy efficiency matching values ​​from high to low and generate a strategy recommendation sequence.

[0029] The strategy execution and feedback module includes a strategy execution unit, a real-time monitoring unit, and a strategy optimization unit. The strategy execution unit is used to select the strategy with the highest energy efficiency matching value from the recommended sequence as the preferred solution. The real-time monitoring unit is used to collect real-time energy consumption data after strategy execution and compare it with the predicted value to calculate the energy consumption deviation rate. The strategy optimization unit is used to set an energy consumption deviation rate threshold, compare the energy consumption deviation with the preset threshold, and determine the effectiveness of the strategy.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] 1. By analyzing historical energy consumption data of equipment from multiple dimensions, and combining multiple indicators such as load rate, energy consumption value, and operating status, the energy intensity index is calculated and a change graph is constructed. Unlike the single total energy consumption analysis in the existing technology, this invention captures the energy efficiency change trend and abnormal fluctuation of equipment in different time periods, and realizes accurate marking and multi-dimensional comprehensive evaluation of historical energy consumption abnormal periods, thereby improving the ability to identify energy consumption anomalies.

[0032] 2. This invention introduces an intelligent candidate strategy generation and evaluation module, which combines historical abnormal period data of equipment and real-time energy consumption demand factors to dynamically generate multiple scheduling strategies and calculate energy efficiency matching values. Unlike the static fixed rule scheduling in the prior art, this invention automatically adjusts the calling order and switching time according to the energy efficiency performance of the equipment and the real-time load to ensure the efficiency and adaptability of the strategy, and at the same time, it quantifies and sorts the strategies through energy efficiency matching values. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating an energy consumption prediction and optimization method based on artificial intelligence according to the present invention.

[0034] Figure 2 This is a schematic diagram of the structure of an energy consumption prediction and optimization system based on artificial intelligence according to the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] In the embodiment: such as Figures 1-2 As shown, the present invention provides a technical solution, an energy consumption prediction and optimization method based on artificial intelligence, the method comprising the following steps:

[0037] Acquire historical data of user-specified commercial facilities and equipment, generate a historical data set, calculate the energy consumption intensity index of each device in different time periods, construct an energy consumption intensity change map, mark abnormal time periods when the historical energy consumption intensity is lower than a preset threshold, and form a set of time periods with abnormal historical energy consumption of equipment.

[0038] Based on the user-specified device information and real-time energy consumption requirements, candidate scheduling strategies are generated. For any strategy, the device calling order and the energy consumption switching time between devices are obtained. When the scheduling time of a device falls into its historical energy consumption abnormal segment set, the actual operating energy consumption of the device during the abnormal time period is calculated.

[0039] The actual total energy consumption and the number of times the predicted energy consumption is exceeded are calculated by traversing the device call time in the strategy. The energy efficiency matching value of each candidate strategy is calculated, and the energy efficiency matching value set is generated and sorted from high to low to obtain the strategy recommendation sequence.

[0040] The strategy with the highest energy efficiency matching value is selected from the recommended strategy sequence as the preferred execution plan. Real-time energy consumption data after execution is collected to calculate the energy consumption deviation rate, and the effectiveness of the strategy is determined by comparing it with the threshold.

[0041] The process involves acquiring historical data of user-specified commercial facilities and equipment, generating a historical data set, calculating the energy consumption intensity index of each device over different time periods, constructing an energy consumption intensity change map, marking abnormal time periods where historical energy consumption intensity is lower than a preset threshold, and forming a set of time segments with abnormal historical energy consumption for the equipment. Specific steps include:

[0042] Retrieve historical data of commercial facilities and equipment specified by the user, and generate a historical data set E = {E1, E2, ..., E...} n}; where E1, E2, ..., E n This represents the historical data of devices 1, 2, ..., n, where n represents the total number of commercial facility devices specified by the user. The historical data information of any selected commercial facility device is represented as E. aHistorical data includes: the load rate of device a in the j-th time period, the energy consumption value of device a in the j-th time period, and the operating status of device a in the j-th time period.

[0043] The energy intensity index of device a in the j-th time period is calculated using the following formula: η aj =C aj / (P aj *T j ); where η aj C represents the energy consumption intensity of device i in the j-th time period. aj P represents the energy consumption of device a in the j-th time period. aj T represents the load rate of device i in the j-th time period. j This represents the duration of the j-th time interval;

[0044] The historical data of the target equipment is traversed to obtain the energy consumption intensity index of different equipment in different time periods. A two-dimensional coordinate system is constructed with each time period as the x-axis and the energy consumption intensity index as the y-axis to form an energy consumption intensity variation map of the equipment. Based on the energy consumption intensity variation map, time periods when the historical energy consumption intensity is lower than a preset threshold are marked to form a set Q of time periods with abnormal historical energy consumption of the equipment. a ={Q a1 Q a2 ,...,Q aM '}, where Q a1 Q a2 ,...,Q aM ' indicates that device a has abnormal energy consumption intensity in the 1st, 2nd, ..., M'th time period, and M' represents the total number of time periods in which abnormal energy consumption efficiency was detected.

[0045] Specifically, in a supermarket, the equipment includes air conditioners (4 units, each with a rated power of 50kW), lighting system (total power of 80kW), freezers (20 units, each with a power of 2kW), and charging piles (5 units, each with a power of 60kW). Peak electricity hours are from 9:00 to 21:00, and off-peak electricity hours are from 21:00 to 9:00 the next day.

[0046] Collect equipment operation data from the supermarket over the past 30 days to generate a historical dataset. For example, for equipment 1: Air Conditioner No. 1: During the time period 8:00-8:30, the load rate is 0.6, the energy consumption is 30kWh, and the operating status is cooling; during the time period 9:00-9:30, the load rate is 0.9, the energy consumption is 45kWh, and the operating status is full-load cooling; during the time period 12:00-12:30, the load rate is 0.7, the energy consumption is 42kWh, and the operating status is cooling.

[0047] Taking 8:00-8:30 as an example, the time period is 0.5 hours. The energy intensity index η is calculated according to the formula. 11 =100kW, with a preset threshold of 80kW. Energy consumption intensity below the threshold is marked as abnormal. For example, if the energy consumption intensity index of device 1 is 120kW from 12:00 to 12:30, it is normal. If the energy consumption intensity index of device 2 is 4kW from 14:00 to 14:30, it is below the threshold and is marked as abnormal.

[0048] Based on user-specified device information and real-time energy consumption requirements, candidate scheduling strategies are generated. For any strategy, the device invocation order and energy consumption switching time between devices are obtained. When the scheduling time of a device falls into its historical energy consumption anomaly segment set, the actual operating energy consumption of the device during the anomaly time segment is calculated. The specific steps include:

[0049] Based on user-specified device information and real-time energy consumption requirements, device scheduling planning is performed, candidate scheduling strategies are generated, and one scheduling strategy is arbitrarily selected and denoted as S. k According to strategy S k The device calling order D(k,i) and the energy consumption switching time between devices are obtained respectively, where T(k,i) represents the time according to strategy S. k The energy adjustment time required to switch from the (i-1)th device to the ith device, D(k,i) represents the time required according to strategy S. k The device called for the i-th time;

[0050] When device D(k,i) is in the time period Located in the historical energy consumption anomaly segment set Q D(k,i) ={Q D(k,i),1 Q D(k,i),2 ,...Q D(k,i),M When the scheduling time falls within a historically inefficient or high-load period of the device, calculate the actual operating energy consumption of device D(k,i) as C. k,i =η D (k,i),j '*P D(k,i),j’ *T j’ ; where η D (k,i),j ' represents the energy consumption intensity of device D(k,i) during the j'-th abnormal time period, P D(k,i),j’ Let T represent the load rate of device D(k,i) during the j'-th abnormal time period, t0 represent the starting time of the policy execution, and T represent the load rate of device D(k,i) during the j'-th abnormal time period. j’ This represents the duration of the j'th abnormal time period.

[0051] Specifically, for real-time load (peak hours 9:00-21:00): air conditioners need to maintain cooling with a load rate greater than or equal to 0.8, freezers need to operate all day with a load rate greater than or equal to 0.9, and charging piles need to prioritize off-peak charging with a peak discharge power less than or equal to 30kW / pile; three candidate strategies are generated based on user-specified equipment information and real-time energy consumption requirements.

[0052] Freezer 2 falls into an abnormal period between 14:00 and 14:30, with an actual energy consumption of 1.4 (normal operating energy consumption is 0.9). Due to the low energy intensity during the abnormal period, the actual energy consumption is higher, and the strategy needs to be adjusted to avoid this.

[0053] The process involves iterating through the device call times in the strategies to calculate the actual total energy consumption and the number of times the predicted energy consumption is exceeded. Then, the energy efficiency matching value for each candidate strategy is calculated, a set of energy efficiency matching values ​​is generated and sorted from high to low, and a strategy recommendation sequence is obtained. Specific steps include:

[0054] Traversal strategy S k Calculate the actual total energy consumption C based on the call time of all devices. k,N And count the number of times K(C) exceeds the predicted energy consumption. k,i >C pred,i ), where C pred,i Indicates device D k,i The predicted energy consumption value, C k,i Indicates device D k,i The actual energy consumption value;

[0055] Computational strategy S k The energy efficiency matching value is defined by the formula shown below: F k =1-K(C k,i >C pred,i ) / N; where K(C k,i >C pred,i The number of times the actual energy consumption of a device exceeds the predicted value in the strategy is represented by , and N represents the total number of devices invoked by the strategy.

[0056] Iterate through all generated candidate policies S1, S2, ..., S R This yields the energy efficiency matching value for each strategy, resulting in a set F = {F1, F2, ..., F...} R}, where R represents the total number of candidate policies generated, F1, F2, ..., F R Let F represent the 1st, 2nd, ..., Rth generated candidate strategies. The obtained set F is sorted from highest to lowest, generating a strategy recommendation sequence F' = {F'1, F'2, ..., F'...}. R}, where F'1>F'2>...>F' R ,F'1,F'2,...,F' RThis represents the 1st, 2nd, ..., Rth candidate strategies generated after sorting.

[0057] Specifically, the actual total energy consumption and the number of times the prediction is exceeded are calculated. For strategy S1, the peak lighting energy consumption is 480kWh (predicted value 500kWh, not exceeded), the charging pile is not running, there is no overspending, and the number of times the prediction is exceeded is 0. For strategy S2, the air conditioner load rate decreases, resulting in insufficient cooling effect. The actual energy consumption exceeds the prediction by 3, and the number of times the prediction is exceeded is 3. For strategy S3, the energy consumption of the freezer during abnormal periods exceeds the prediction by 1, and the number of times the prediction is exceeded is 1. The energy efficiency matching value of each strategy is calculated. Based on the calculated energy efficiency matching value of each strategy, the generated candidate strategies are sorted from high to low to generate a strategy recommendation sequence.

[0058] The strategy with the highest energy efficiency matching value in the recommended strategy sequence is selected as the preferred execution plan. Real-time energy consumption data after execution is collected to calculate the energy consumption deviation rate, which is then compared with a threshold to determine the effectiveness of the strategy. The specific steps include:

[0059] The strategy with the highest energy efficiency matching value is selected from the recommended strategy sequence as the preferred execution plan;

[0060] Real-time energy consumption data after the strategy is executed is collected and compared with the predicted value to calculate the energy consumption deviation rate, defined as follows: Z = |C real -C pred | / C pred *100%; of which, C real C represents the real-time energy consumption data after the strategy is executed. pred represents the predicted energy consumption value, and Z represents the energy consumption deviation rate;

[0061] Set an energy consumption deviation rate threshold Z0. When Z≤Z0, the strategy is deemed effective and the current strategy is maintained. When Z>Z0, the strategy is deemed not to have met expectations, triggering a strategy switching mechanism. Select the second-best scoring strategy F'2 from the recommended strategy sequence and update it as the current execution plan. Repeat the monitoring and switching process until either of the following conditions is met: energy consumption deviation rate threshold Z≤Z0, strategy execution is successful; all strategies have been tried and failed to meet the target, triggering an alarm.

[0062] Specifically, the preferred strategy S1 is executed, real-time energy consumption data after the strategy is executed is collected and compared with the predicted value, the energy consumption deviation rate is calculated to be 1.05%, and the energy consumption deviation rate threshold is set to 5%. At this time, if the energy consumption deviation rate is less than or equal to the energy consumption deviation rate threshold, the strategy is determined to be effective and execution is maintained.

[0063] An AI-based energy consumption prediction and optimization system is applied to the AI-based energy consumption prediction and optimization method described in any one of claims 1-5. The system includes: a data acquisition module, a candidate strategy generation module, a strategy evaluation and ranking module, and a strategy execution and feedback module. The data acquisition module calculates the energy consumption intensity index of each device in different time periods, forming a set of historical energy consumption anomalies for each device. The candidate strategy generation module generates candidate scheduling strategies. For any strategy, it obtains the device calling order and the energy consumption switching time between devices. When the scheduling time of a device falls into its historical energy consumption anomaly set, it calculates the actual operating energy consumption of the device during the anomaly time period. The strategy evaluation and ranking module iterates through the device calling times in the strategies, calculates the actual total energy consumption and the number of times the predicted energy consumption is exceeded, calculates the energy efficiency matching value of each candidate strategy, and generates a strategy recommendation sequence. The strategy execution and feedback module selects the strategy with the highest energy efficiency matching value from the strategy recommendation sequence as the preferred execution plan, collects real-time energy consumption data after execution, calculates the energy consumption deviation rate, and compares it with a threshold to determine the effectiveness of the strategy.

[0064] The data acquisition module includes a data acquisition unit, an energy consumption intensity calculation unit, and an abnormal time period marking unit. The data acquisition unit is used to acquire historical data of user-specified devices and generate a historical data set. The energy consumption intensity calculation unit is used to calculate the energy consumption intensity index of each device in different time periods using formulas. The abnormal time period marking unit is used to construct an energy consumption intensity change graph, compare it with a preset threshold, mark the abnormal time periods when the energy consumption intensity is lower than the threshold, and generate an abnormal time period set.

[0065] The candidate strategy generation module includes a strategy planning unit and an abnormal energy consumption calculation unit. The strategy planning unit is used to generate candidate scheduling strategies based on the device information and real-time energy consumption requirements specified by the user. Each strategy includes the device call order and switching time. The abnormal energy consumption calculation unit is used to calculate the actual operating energy consumption when the device scheduling time in the strategy falls into its abnormal time segment set.

[0066] The strategy evaluation and ranking module includes an energy consumption statistics unit, an energy efficiency matching calculation unit, and a strategy ranking unit. The energy consumption statistics unit is used to traverse the call time of all devices in the strategy, calculate the actual total energy consumption, and count the number of times the actual energy consumption exceeds the predicted value. The energy efficiency matching calculation unit is used to calculate the energy efficiency matching value of the strategy. The strategy ranking unit is used to sort the set of energy efficiency matching values ​​from high to low and generate a strategy recommendation sequence.

[0067] The strategy execution and feedback module includes a strategy execution unit, a real-time monitoring unit, and a strategy optimization unit. The strategy execution unit selects the strategy with the highest energy efficiency matching value from the recommended sequence as the preferred solution. The real-time monitoring unit collects real-time energy consumption data after strategy execution and compares it with the predicted value to calculate the energy consumption deviation rate. The strategy optimization unit sets the energy consumption deviation rate threshold, compares the energy consumption deviation with the preset threshold, and determines the effectiveness of the strategy.

[0068] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. An energy consumption prediction and optimization method based on artificial intelligence, characterized in that: The method includes the following steps: Acquire historical data of user-specified commercial facilities and equipment, generate a historical data set, calculate the energy consumption intensity index of each device in different time periods, construct an energy consumption intensity change map, mark abnormal time periods when the historical energy consumption intensity is lower than a preset threshold, and form a set of time periods with abnormal historical energy consumption of equipment. Based on the user-specified device information and real-time energy consumption requirements, candidate scheduling strategies are generated. For any strategy, the device calling order and the energy consumption switching time between devices are obtained. When the scheduling time of a device falls into its historical energy consumption abnormal segment set, the actual operating energy consumption of the device during the abnormal time period is calculated. The actual total energy consumption and the number of times the predicted energy consumption is exceeded are calculated by traversing the device call time in the strategy. The energy efficiency matching value of each candidate strategy is calculated, and the energy efficiency matching value set is generated and sorted from high to low to obtain the strategy recommendation sequence. Traversal strategy S k Calculate the actual total energy consumption C based on the call time of all devices. k,N And count the number of times K(C) exceeds the predicted energy consumption. k,i >C pred,i ), where C pred,i Indicates device D k,i The predicted energy consumption value, C k,i Indicates device D k,i The actual energy consumption value; Computational strategy S k The energy efficiency matching value is defined by the formula shown below: F k =1-K(C k,i >C pred,i ) / N; where K(C k,i >C pred,i The number of times the actual energy consumption of a device exceeds the predicted value in the strategy is represented by , and N represents the total number of devices invoked by the strategy. Iterate through all generated candidate policies S1, S2, ..., S R This yields the energy efficiency matching value for each strategy, resulting in a set F = {F1, F2, ..., F...} R }, where R represents the total number of candidate policies generated, F1, F2, ..., F R Let F' represent the 1st, 2nd, ..., Rth generated candidate strategies. The obtained set F is sorted from highest to lowest, generating a strategy recommendation sequence F' = {F'1, F'2, ..., F'...}. R }, where F'1>F'2>...>F' R ,F'1,F'2,...,F' R This represents the 1st, 2nd, ..., Rth candidate strategies generated after sorting; The strategy with the highest energy efficiency matching value is selected from the recommended strategy sequence as the preferred execution plan. Real-time energy consumption data after execution is collected to calculate the energy consumption deviation rate, and the effectiveness of the strategy is determined by comparing it with the threshold.

2. The energy consumption prediction and optimization method based on artificial intelligence according to claim 1, characterized in that: The process involves acquiring historical data of user-specified commercial facilities and equipment, generating a historical data set, calculating the energy consumption intensity index of each device over different time periods, constructing an energy consumption intensity change map, marking abnormal time periods where historical energy consumption intensity is lower than a preset threshold, and forming a set of time segments with abnormal historical energy consumption for the equipment. Specific steps include: Retrieve historical data of user-specified commercial facilities and equipment, and generate a historical data set E={E1,E2,...,E...} n }; where E1, E2, ..., E n This represents the historical data of devices 1, 2, ..., n, where n represents the total number of commercial facility devices specified by the user. The historical data information of any selected commercial facility device is represented as E. a The historical data information includes: the load rate of device a in the j-th time period, the energy consumption value of device a in the j-th time period, and the operating status of device a in the j-th time period. The energy intensity index of device a in the j-th time period is calculated using the following formula: η aj =C aj / (P aj *T j ); where η aj C represents the energy consumption intensity of device i in the j-th time period. aj P represents the energy consumption of device a in the j-th time period. aj T represents the load rate of device i in the j-th time period. j This represents the duration of the j-th time interval; The historical data of the target equipment is traversed to obtain the energy consumption intensity index of different equipment in different time periods. A two-dimensional coordinate system is constructed with each time period as the x-axis and the energy consumption intensity index as the y-axis to form an energy consumption intensity variation map of the equipment. Based on the energy consumption intensity variation map, time periods when the historical energy consumption intensity is lower than a preset threshold are marked to form a set Q of time periods with abnormal historical energy consumption of the equipment. a ={Q a1 Q a2 ,...,Q aM’ }, where Q a1 Q a2 ,...,Q aM’ This indicates that device a has abnormal energy consumption intensity during the 1st, 2nd, ..., M'th time periods, where M' represents the total number of time periods during which abnormal energy consumption efficiency was detected.

3. The energy consumption prediction and optimization method based on artificial intelligence according to claim 2, characterized in that: Based on user-specified device information and real-time energy consumption requirements, candidate scheduling strategies are generated. For any strategy, the device invocation order and energy consumption switching time between devices are obtained. When the scheduling time of a device falls into its historical energy consumption anomaly segment set, the actual operating energy consumption of the device during the anomaly time segment is calculated. The specific steps include: Based on user-specified device information and real-time energy consumption requirements, device scheduling planning is performed, candidate scheduling strategies are generated, and one scheduling strategy is arbitrarily selected and denoted as S. k According to strategy S k The device calling order D(k,i) and the energy switching time between devices T(k,i) are obtained respectively, where T(k,i) represents the time according to strategy S. k The energy adjustment time required to switch from the (i-1)th device to the ith device, D(k,i) represents the time required according to strategy S. k The device called for the i-th time; When device D(k,i) is in the time period Located in the historical energy consumption anomaly segment set Q D(k,i) ={Q D(k,i),1 Q D(k,i),2 ,...Q D(k,i),M’ When}, the actual operating energy consumption of computing device D(k,i) is C. k,i =η D(k,i),j’ *P D(k,i),j’ *T j’ ; where η D(k,i),j’ P represents the energy consumption intensity of device D(k,i) during the j'-th abnormal time period. D(k,i),j’ Let T represent the load rate of device D(k,i) during the j'-th abnormal time period, t0 represent the starting time of the policy execution, and T represent the load rate of device D(k,i) during the j'-th abnormal time period. j’ This represents the duration of the j'th abnormal time period.

4. The energy consumption prediction and optimization method based on artificial intelligence according to claim 3, characterized in that: The strategy with the highest energy efficiency matching value in the recommended strategy sequence is selected as the preferred execution plan. Real-time energy consumption data after execution is collected to calculate the energy consumption deviation rate, which is then compared with a threshold to determine the effectiveness of the strategy. The specific steps include: The strategy with the highest energy efficiency matching value is selected from the recommended strategy sequence as the preferred execution plan; Real-time energy consumption data after the strategy is executed is collected and compared with the predicted value to calculate the energy consumption deviation rate, defined as follows: Z=|C real -C pred | / C pred *100%; of which, C real C represents the real-time energy consumption data after the strategy is executed. pred represents the predicted energy consumption value, and Z represents the energy consumption deviation rate; A threshold Z0 for energy consumption deviation rate is set. When Z≤Z0, the strategy is deemed effective and the current strategy is maintained. When Z>Z0, the strategy is deemed not to have met expectations, and a strategy switching mechanism is triggered. The second-best scoring strategy F'2 is selected from the recommended strategy sequence and updated as the current execution plan. The monitoring and switching process is repeated until either of the following conditions is met: energy consumption deviation rate threshold Z≤Z0, the strategy is executed successfully; all strategies have been tried and have failed to meet the target, an alarm is triggered.

5. An artificial intelligence-based energy consumption prediction and optimization system, applied to the artificial intelligence-based energy consumption prediction and optimization method described in any one of claims 1-4, characterized in that: The system includes: a data acquisition module, a candidate strategy generation module, a strategy evaluation and ranking module, and a strategy execution and feedback module. The data acquisition module calculates the energy consumption intensity index of each device in different time periods, forming a set of historical energy consumption anomalies for each device. The candidate strategy generation module generates candidate scheduling strategies. For any strategy, it obtains the device calling order and the energy consumption switching time between devices. When the scheduling time of a device falls into its historical energy consumption anomaly set, it calculates the actual operating energy consumption of the device during the anomaly period. The strategy evaluation and ranking module iterates through the device calling times in the strategies, calculates the actual total energy consumption and the number of times the predicted energy consumption is exceeded, calculates the energy efficiency matching value of each candidate strategy, and generates a strategy recommendation sequence. The strategy execution and feedback module selects the strategy with the highest energy efficiency matching value from the strategy recommendation sequence as the preferred execution plan, collects real-time energy consumption data after execution, calculates the energy consumption deviation rate, and compares it with a threshold to determine the effectiveness of the strategy.

6. The energy consumption prediction and optimization system based on artificial intelligence according to claim 5, characterized in that: The data acquisition module includes a data acquisition unit, an energy consumption intensity calculation unit, and an abnormal time period marking unit. The data acquisition unit is used to acquire historical data of user-specified devices and generate a historical data set. The energy consumption intensity calculation unit is used to calculate the energy consumption intensity index of each device in different time periods using a formula. The abnormal time period marking unit is used to construct an energy consumption intensity change graph, compare it with a preset threshold, mark the abnormal time periods when the energy consumption intensity is lower than the threshold, and generate an abnormal time period set.

7. The energy consumption prediction and optimization system based on artificial intelligence according to claim 6, characterized in that: The candidate strategy generation module includes a strategy planning unit and an abnormal energy consumption calculation unit. The strategy planning unit is used to generate candidate scheduling strategies based on the device information and real-time energy consumption requirements specified by the user. Each strategy includes the device calling order and switching time. The abnormal energy consumption calculation unit is used to calculate the actual operating energy consumption when the device scheduling time in the strategy falls into its abnormal time segment set.

8. The energy consumption prediction and optimization system based on artificial intelligence according to claim 7, characterized in that: The strategy evaluation and ranking module includes an energy consumption statistics unit, an energy efficiency matching calculation unit, and a strategy ranking unit. The energy consumption statistics unit is used to traverse the call time of all devices in the strategy, calculate the actual total energy consumption, and count the number of times the actual energy consumption exceeds the predicted value. The energy efficiency matching calculation unit is used to calculate the energy efficiency matching value of the strategy; the strategy sorting unit is used to sort the set of energy efficiency matching values ​​from high to low and generate a strategy recommendation sequence.

9. The energy consumption prediction and optimization system based on artificial intelligence according to claim 8, characterized in that: The strategy execution and feedback module includes a strategy execution unit, a real-time monitoring unit, and a strategy optimization unit. The strategy execution unit is used to select the strategy with the highest energy efficiency matching value from the recommended sequence as the preferred solution. The real-time monitoring unit is used to collect real-time energy consumption data after the strategy is executed, compare it with the predicted value, and calculate the energy consumption deviation rate; the strategy optimization unit is used to set the energy consumption deviation rate threshold, compare the energy consumption deviation with the preset threshold, and determine the effectiveness of the strategy.

Images