Heating energy system optimization scheduling method based on big data processing

Abstract

The invention relates to the technical field of big data service, and provides a heating energy system optimization scheduling method based on big data processing, and the method comprises the following steps: S1, collecting various data in a heating energy system, carrying out the preprocessing of the data, and constructing a big database of the heating energy system; s2, based on the preprocessed data, combining historical data to train a thermal load prediction model, and utilizing a big data analysis technology to mine potential factors affecting thermal load changes; and S3, taking the load prediction result as known input, considering heat loss in a heat supply network transmission process, and establishing an optimal scheduling model of the heating energy system. Potential factors influencing thermal load change are mined by using a big data analysis technology, and the abundant information is taken as model input characteristics, so that the change trend of the thermal load can be captured more comprehensively and accurately, the prediction precision can be remarkably improved, the energy waste can be reduced, and the energy utilization efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the field of big data service technology, and in particular to a heating energy system optimization scheduling method based on big data processing. Background Art

[0002] Under the background of global advocacy of energy conservation, emission reduction and sustainable development, the heating energy system is an important area of ​​energy consumption, and its optimized scheduling is of great significance to improving energy utilization efficiency, reducing operating costs and reducing environmental pollution. However, the traditional heating energy scheduling methods often only consider a single or a few factors, and cannot accurately capture the dynamic changes of heat load. In practical applications, the prediction results of these methods deviate greatly from the actual heat load demand, resulting in a mismatch between energy supply and demand, causing energy waste or insufficient supply. Previous scheduling models usually treat these uncertainties as deterministic factors, or simply use a fixed safety margin to deal with them. This method is too conservative, which not only increases the operating cost of the system, but also cannot effectively deal with extreme situations. Once the system fails or the operation deviates from the optimal state, it often takes a long time to recover, resulting in energy waste and economic losses. Summary of the invention

[0003] In view of the problems existing in the prior art, the purpose of the present invention is to provide a heating energy system optimization scheduling method based on big data processing to solve the problems raised by the above background technology.

[0004] To solve the above problems, the present invention provides a heating energy system optimization scheduling method based on big data processing, comprising the following steps:

[0005] S1. Collect various data in the heating energy system, pre-process the data, and build a large database of the heating energy system;

[0006] S2. Based on the preprocessed data, the heat load prediction model is trained in combination with historical data, and the potential factors affecting the heat load change are explored using big data analysis technology;

[0007] S3, taking the load forecast results as known inputs and considering the heat loss during the heat network transmission process, an optimal scheduling model for the heating energy system is established;

[0008] S4. Use probability distribution estimation methods based on big data to quantitatively analyze uncertainty factors, build opportunity constraints, and reduce system operation risks;

[0009] S5. Use intelligent optimization algorithm to solve the optimization scheduling model and generate detailed intelligent optimization strategy according to the solution results;

[0010] S6. Monitor the actual operating status of the real-time monitoring system, and when a deviation occurs, promptly activate the dynamic adjustment mechanism to ensure that the system is always in the optimal operating state.

[0011] Preferably, in step S1, constructing the big database of the heating energy system includes the following steps:

[0012] S11. Use various sensors distributed in the heating energy system to collect real-time meteorological data, heat source equipment operation data, heat network operation data, and user heat load data;

[0013] S12. Conduct anomaly detection on the collected data, correct the abnormal data, and the calculation formula is In the formula, μ is the moving window mean, σ is the moving window standard deviation, and impute the missing values to ensure data integrity. The calculation formula is In the formula, x t is the value of the variable at time t, is the value of the variable at t + i, and NN is the predicted value obtained through the neural network algorithm;

[0014] S13. Unify the data format, add timestamps and labels to the data, and establish a time series database;

[0015] S14. Verify the data integrity and verify the data consistency;

[0016] S15. Add label indexes to the data and store the data distributively.

[0017] Preferably, in step S2, mining the potential factors affecting the heat load change includes the following steps:

[0018] S21. Statistically analyze the heat load change rules in different time periods, analyze the influence of residents' work and rest time on the heat load, and calculate the average heat load in each time period. The calculation formula is In the formula, is the average heat load, is the heat load at the t-th moment in the i-th time period, and n is the number of data in this time period;

[0019] S22. Use the meteorological data, date type, time information, and the mined potential influencing factors as the input features of the model;

[0020] S23. Divide the model input features into a training set, a validation set, and a test set according to the ratio of 7:2:1, and use random forest regression to train on randomly selected feature subsets. The calculation formula is In the formula, is the predicted value after training, m is the number of decision trees, and y j is the predicted value of the j-th decision tree;

[0021] S24. Evaluate the trained model using the validation set and the test set. The calculation formula is In the formula, MSE is the evaluation result, y i is the true value, is the predicted value, and n is the number of samples.

[0022] Preferably, in step S3, establishing the optimal scheduling model of the heating energy system includes the following steps:

[0023] S31. Taking the minimization of the system operation cost, the maximization of the energy utilization rate, and the minimization of carbon emissions as the comprehensive goal, construct a multi-objective optimization model to lay the foundation for the optimal scheduling of the entire heating energy system. The calculation formula for the minimization of the system operation cost is In the formula, C t is the total system operation cost, C f is the fuel cost, C m is the equipment maintenance cost, C p is the energy purchase cost from the external power grid or gas network. T is the total number of time periods. i ∈ CHP is the set of numbers of combined heat and power equipment. a i is the fuel cost coefficient per unit power generation of combined heat and power equipment i, is the power generation power of combined heat and power equipment i at time t, b i is the fuel cost coefficient per unit heat supply of combined heat and power equipment i, is the power generation power of combined heat and power equipment i at time t. j ∈ GB is the set of numbers of gas filtering equipment. c j is the fuel cost coefficient per unit heat supply of gas boiler equipment j, is the heat supply power of gas boiler equipment j at time t. k ∈ equip is the set of numbers of all equipment. d k is the unit power maintenance cost coefficient of equipment k, P k,t is the operating power of equipment k at time t, λ grid,t 、P grid,t 、λ gas,t and G gas,t are respectively the electricity price for purchasing electricity from the power grid, the purchased electric power from the power grid, the gas price for purchasing gas from the gas network, and the purchased gas volume from the gas network at time t. The calculation formula for the maximization of the energy utilization rate is In the formula, η t is the total energy utilization rate of the system, and E input,t are respectively the electricity load demand of the system, the heat load demand of the system, and the total energy input to the system at time t, ηconv For the energy conversion efficiency, the calculation formula for minimizing carbon emissions is In the formula, CO2 is the total carbon emissions of the system, and γ CHP and γ GB are respectively the carbon emission coefficients per unit power generation of the combined heat and power equipment and the carbon emission coefficients per unit heat supply of the gas boiler equipment;

[0024] S32. Establish the conditions of energy balance constraint, equipment operation constraint and heat network transmission loss constraint, so that the system can stably meet the load demand of users, prevent equipment damage caused by over - operation or power mutation, and conform to the actual system operation situation;

[0025] S33. Select the improved non - dominated sorting genetic algorithm combined with the multi - objective optimization model and constraint conditions. Through coding and initialization, convert the decision variables into chromosomes, and the fitness calculation can evaluate the advantages and disadvantages of each individual under multiple objectives;

[0026] S34. Output the optimal scheduling plan according to the multi - objective optimization model and energy constraint conditions.

[0027] Preferably, in step S32, the energy balance constraint conditions are divided into power balance constraint and heat balance constraint. The calculation formula for the power balance constraint is In the formula, P grid,t 、P solar,t 、 and P ET,t are respectively the power generation power of the combined heat and power equipment, the electric power purchased from the power grid, the power generation power of solar energy, the power load demand of the system and the power consumption of the electric boiler at time t. The calculation formula for the heat balance constraint is In the formula, and are respectively the heat supply power of the combined heat and power equipment, the heat supply power of the gas boiler, the heat release power of the heat storage device, the heat load demand of the system and the heat charging power of the heat storage device at time t. The equipment operation constraint conditions are divided into power limit constraint and ramp rate limit constraint. The calculation formula for the power limit constraint is In the formula, and are respectively the minimum power generation power and the maximum power generation power of the combined heat and power equipment i, and are respectively the minimum heat supply power and the maximum heat supply power of the gas boiler equipment j. The calculation formula for the ramp rate limit constraint is In the formula, R down and R upThey are the maximum descending rate and the maximum ascending rate of the device's power generation respectively. The calculation formula for the heat network transmission loss constraint condition is In the formula,[[]]END]] and are the heat actually delivered to the user end by the heat network and the total heat generated by the heat source at time t respectively, and α loss is the heat network transmission loss rate.

[0028] Preferably, in step S33, the fitness calculation for evaluating the advantages and disadvantages of each individual under multiple objectives includes the following steps:

[0029] S331. Perform non-dominated sorting on all individuals in the population, divide the individuals into different non-dominated levels, and calculate their crowding distance. The calculation formula is In the formula, X i+1 and X i-1 are the next individual and the previous individual of the non-dominated individual X i respectively, and d i is the crowding distance;

[0030] S332. Randomly select pairs of parent individuals from the parent individuals obtained by the selection operation to generate offspring individuals. The calculation formula is In the formula, X a and X b are each pair of parent individuals, and X c and X d are each pair of offspring individuals

[0031] S333. For the offspring individuals obtained after crossover, determine the individuals that need to mutate with a mutation probability, and calculate the mutated components through polynomial mutation. The calculation formula is In the formula, x′ ij is the new value of the j-th component of the individual X i after polynomial mutation operation, x ij is the original value of the j-th component of the individual X i that needs to mutate, r is a random number uniformly distributed in the interval [0 - 1], and Δ1 and Δ2 are intermediate variables for calculating the mutation amount, and Among them, and are the upper and lower limits of the value range of x ij respectively, and η m is the mutation distribution index;

[0032] S334. Combine the parent population and the offspring population after crossover and mutation into a combined population. The calculation formula is R t = P t ∪ Q t , in the formula, R tFor merging populations, P t is the parental population, Q t is the offspring population. Perform non - dominated sorting on the merged population, and successively select individuals in the non - dominated levels to join the new population P t+1 until the number of individuals in the new population reaches the population size N;

[0033] S335. Check whether the current iteration number t reaches the preset maximum iteration number T max and calculate the difference between the non - dominated solution sets obtained from two adjacent iterations. When t ≥ T max and the difference is less than the preset threshold ∈, the algorithm terminates, and the non - dominated solutions in the current population are output as the optimization results.

[0034] Preferably, in step S4, constructing chance constraints and reducing the system operation risk includes the following steps:

[0035] S41. Collect data of historical temperature, humidity, and wind speed, and calculate the temperature correction coefficient. The calculation formula is In the formula, α temp is the temperature correction coefficient, T actual is the actual temperature, T base is the loss temperature. Statistically analyze the historical heat load curve and extract the peak - valley characteristics. The calculation formula is ΔQ pv = Q max - Q min . In the formula, ΔQ pv is the peak - valley heat load difference, Q max and Q min are the maximum and minimum values of the heat load respectively, and calculate the prediction error rate. The calculation formula is In the formula, ∈ t is the prediction error rate, Q f,t and Q a,t are the predicted value and the actual value of the heat load at time t respectively;

[0036] S42. Analyze the distribution characteristics of heat load data, meteorological data, etc. The calculation formula is In the formula, f(x) is the probability density value at the data point x, n is the number of samples, h is the bandwidth parameter, x i is the i - th sample data point, K(·) is the Gaussian kernel function. Construct a joint probability map model of meteorology and load, and quantify the uncertainty through the posterior probability P(Q load |T, W). In the formula, T is the temperature and W is the humidity;

[0037] D43. Use the gradient fuzzy number algorithm to describe the parameters with uncertainty. The calculation formula is where μ(x) is the membership function of the independent variable x with respect to the given trapezoidal fuzzy set, and a, b, c, and d are all parameters of the trapezoidal fuzzy number;

[0038] S44. Establish an opportunity constraint transformation to ensure that the power supply for heating energy meets the load demand. The opportunity constraint transformation includes power balance constraints and equivalent crisp constraints, and the calculation formula is where Pr(·) is the probability operation, and ∑ i P i,t is the sum of the power generation of all power generation equipment at time t, β is the confidence level, δ is the coefficient related to the confidence level β, and δ = Φ -1 (1 - β), where Φ - (·) is the inverse cumulative distribution function of the standard normal distribution, and σ load is the standard deviation of the power load demand;

[0039] S45. Simulate uncertain factors through random sampling and construct a two-stage robust optimization model. Considering the uncertainty, minimize the total cost, and the calculation formula is where is the load value under the s-th load scenario at time t, is the mean μ t , variance of the normal distribution, min x is the minimization operation of the variable x, is the expectation operation, C(x, ζ) is the cost function, ρ is the risk aversion coefficient, with a value range of [0 - 1], CVaR α (·) is the conditional value at risk, α is the confidence level, with a value range of [0 - 1], is the scenario probability π s multiplied by the sum of the adjustment amounts at time t under all scenarios s, is the scenario probability π s multiplied by the sum of the load values at time t under all scenarios s;

[0040] S46. In the scenario of dynamic update of uncertainty parameters, optimize the model parameters in real time according to the continuously obtained new data, and dynamically adjust the threshold according to the fluctuation of the load data. The calculation formula is where P(θ|D 1:t ) is the posterior probability distribution of the model parameter θ obtained from the historical data set D 1:t from time 1 to time t, P(D t |θ) is the likelihood function, β t is the dynamic threshold at time t, β0 is the initial threshold, γ is the adjustment factor, and Var(Q load,t ) is the load Q at time tload,t Variance;

[0041] S47. During the time period, minimize the sum of the fuel cost and the reserve cost to ensure that the total power generation of the power generation equipment can meet the load demand. The calculation formula is min∑ t (C f,t +C r,t ), where C f,t is the fuel cost at time t, and C r,t is the reserve cost at time t.

[0042] Preferably, in the step S5, the generated adjustment strategy is divided into heat source output adjustment and heat network dynamic regulation. The priority order of the heat source output adjustment is to start the standby boiler, adjust the CHP unit, and call the energy storage. The calculation formula is where P GB,t , P CHP,t and are the power generation of the standby boiler, the power generation of the CHP unit, and the discharge of the energy storage system at time t respectively, P GB,max is the maximum power generation of the standby boiler, ΔQ and ΔQ GB,max are the change in heat load and the maximum heat supply capacity of the standby boiler respectively, is the smaller value of 1 and to prevent exceeding its maximum power generation, R up is the rising power regulation rate of the CHP unit, Δt is the time interval, Q ST,state is the current state of charge of the energy storage system, η dis is the discharge efficiency of the energy storage system, is the smaller value of Q ST,state and to ensure that the discharge of the energy storage system does not exceed its remaining charge. The heat network dynamic regulation is controlled by adjusting the valve opening. The calculation formula is where α v,t is the opening of the heat network valve at time t, K p is the proportionality coefficient with a value of 0.5, ΔT t is the temperature deviation at time t, K i is the integral coefficient with a value of 0.1, ∫ΔT t dt is the integral value of the temperature deviation over time, K d is the differential coefficient with a value of 0.2, is the rate of change of the temperature deviation.

[0043] Preferably, in the step S6, when a deviation occurs, starting the dynamic adjustment mechanism includes the following steps:

[0044] S61. Set the trigger threshold for starting the dynamic adjustment mechanism, and the trigger condition is where L is the emergency adjustment, N is the normal adjustment, and ΔP t is the power deviation, and ΔT t is the temperature deviation, and calculate the deviation in real time. The calculation formula is where P a,t and P s,t are the actual power and the planned power at time t respectively, and T u,t and T s,t are the supply temperature and the actual temperature at time t;

[0045] S62. By inputting the feature vector at the current moment and the updated model parameters, obtain the load prediction value at the next moment. The calculation formula is where is the load prediction value at the next moment, M(·) is the long short-term memory network model, and X t is the feature vector at time t, and θ new are the updated LSTM model parameters;

[0046] S63. Evaluate the heating optimization adjustment performance, which reflects the correction degree of the adjustment strategy to the heat load at different moments. The calculation formula is where ξ is the adjustment efficiency, and ΔQ c,t is the change in the heat load after adjustment at time t, and ΔQ i,t is the initial change in the heat load at time t, ΔC is the change in cost, and C a is the cost after adjustment, and C o is the original cost;

[0047] S64. Establish a feedback mechanism to correct the conversion efficiency parameter. The calculation formula is where η conv,new and η conv,old are the conversion efficiency parameters after and before correction respectively, MSE load is the mean square error of the load prediction, and Q mean is the load mean;

[0048] S65. Use the transfer learning method, combine the basic model and the real-time data, and obtain the final model parameters. The calculation formula is θ final = TransferLearning(f(x), x s ), where θ final are the final model parameters, TransferLearning(·) is the transfer learning algorithm, f(x) is the basic model, and x s is the real-time data.

[0049] Preferably, in step S11, the sensors include but are not limited to temperature sensors, pressure sensors, flow sensors, and power sensors, and the data acquisition frequency is dynamically adjusted according to different data types and system requirements to ensure that the collected data can accurately reflect the real-time operating status of the system.

[0050] The invention provides a heating energy system optimization scheduling method based on big data processing, which has the following beneficial effects:

[0051] 1. The present invention is based on a large amount of preprocessed data, combined with historical data to train a heat load prediction model, and uses big data analysis technology to explore potential factors that affect heat load changes, such as residents' living habits and industrial production activities. By using this rich information as model input features, it can capture the changing trend of heat load more comprehensively and accurately, significantly improve prediction accuracy, effectively reduce the mismatch between energy supply and demand, reduce energy waste, and improve energy utilization efficiency.

[0052] 2. By using the probability distribution estimation method based on big data, the uncertainty factors of meteorological data and heat load forecasts are quantitatively analyzed, and opportunity constraints are constructed to incorporate them into the optimization scheduling model, so that the model can generate a more robust scheduling plan under the condition of meeting a certain confidence level, greatly reducing the system operation risk. With the help of big data technology to explore potential influencing factors, we can have a more comprehensive understanding of the driving factors of heat load changes, thereby achieving more refined energy scheduling. After considering these potential factors, the system can more reasonably arrange the power generation and heating power of heat source equipment, optimize the energy allocation plan, improve energy utilization efficiency, and reduce system operating costs.

[0053] 3. By re-forecasting loads and optimizing scheduling calculations based on the latest monitoring data, the scheduling plan can be adjusted online to ensure that the system is always in the optimal operating state. This real-time monitoring and dynamic adjustment mechanism can quickly respond to changes in the system and improve the reliability and stability of the system. At the same time, by adjusting the opening of each regulating valve in the heating network in real time, the economy and stability of the heating network operation can be ensured, and the operating efficiency of the entire heating energy system can be further improved. It can also be quickly adjusted according to real-time data and dynamic changes, making the heating energy system more adaptable and flexible. Whether it is responding to sudden changes in weather, fluctuations in user heat load demand, or equipment failures and other emergencies, the system can respond in a timely manner, adjust the scheduling strategy, and ensure the quality of heating services, which is conducive to improving residents' heating satisfaction and also provides strong support for the sustainable management of urban energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 A schematic diagram of the process steps of a heating energy system optimization scheduling method based on big data processing provided in this application. DETAILED DESCRIPTION

[0056] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples of the specification. The following examples are only used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0057] like Figure 1 As shown, this embodiment proposes a heating energy system optimization scheduling method based on big data processing, including the following steps:

[0058] S1. Collect various data in the heating energy system, pre-process the data, and build a large database of the heating energy system;

[0059] S2. Based on the preprocessed data, the heat load prediction model is trained in combination with historical data, and the potential factors affecting the heat load change are explored using big data analysis technology;

[0060] S3, taking the load forecast results as known inputs and considering the heat loss during the heat network transmission process, an optimal scheduling model for the heating energy system is established;

[0061] S4. Use probability distribution estimation methods based on big data to quantitatively analyze uncertainty factors, build opportunity constraints, and reduce system operation risks;

[0062] S5. Use intelligent optimization algorithm to solve the optimization scheduling model and generate detailed intelligent optimization strategy according to the solution results;

[0063] S6. Monitor the actual operating status of the system in real time, and when deviations occur, promptly activate the dynamic adjustment mechanism to ensure that the system is always in the optimal operating state.

[0064] In this embodiment, in step S1, building a big database of the heating energy system includes the following steps:

[0065] S11. Use various sensors distributed in the heating energy system to collect real-time meteorological data, heat source equipment operation data, heat network operation data and user heat load data;

[0066] S12. Perform anomaly detection on the collected data, correct the abnormal data, and the calculation formula is In the formula, μ is the moving window mean, σ is the moving window standard deviation, and impute the missing values to ensure data integrity. The calculation formula is In the formula, x t is the value of the variable at time t, is the value of the variable at t + i, and NN is the predicted value obtained through the neural network algorithm;

[0067] S13. Unify the data format, add timestamps and labels to the data, and establish a time series database;

[0068] S14. Verify the data integrity and validate the data consistency;

[0069] S15. Add label indexes to the data and store the data distributively.

[0070] In this embodiment, in step S2, the steps for mining the potential factors affecting the change of heat load include the following:

[0071] S21. Statistically analyze the change law of heat load in different time periods, analyze the influence of residents' work and rest time on heat load, and calculate the average heat load for each time period. The calculation formula is In the formula, is the average heat load, is the heat load at the t-th moment in the i-th time period, and n is the number of data in this time period;

[0072] S22. Use the meteorological data, date type, time information, and the mined potential influencing factors as the input features of the model;

[0073] S23. Divide the model input features into a training set, a validation set, and a test set according to the ratio of 7:2:1, and use random forest regression to train on randomly selected feature subsets. The calculation formula is In the formula, is the predicted value after training, m is the number of decision trees, and y j is the predicted value of the j-th decision tree;

[0074] S24. Use the validation set and the test set to evaluate the trained model. The calculation formula is In the formula, MSE is the evaluation result, y i is the true value, is the predicted value, and n is the number of samples.

[0075] Specifically, based on a large amount of preprocessed data, the present invention trains a heat load prediction model in combination with historical data, and uses big data analysis technology to mine potential factors affecting heat load changes, such as residents' living habits and industrial production activity rules. By using this rich information as input features of the model, the changing trend of heat load can be captured more comprehensively and accurately. Compared with traditional prediction methods, it can significantly improve the prediction accuracy, effectively reduce the mismatch between energy supply and demand, reduce energy waste, and improve energy utilization efficiency.

[0076] In this embodiment, in step S3, establishing an optimal scheduling model for the heating energy system includes the following steps:

[0077] S31. Taking the minimization of system operation cost, the maximization of energy utilization rate, and the minimization of carbon emissions as comprehensive objectives, constructing a multi-objective optimization model to lay a foundation for the optimal scheduling of the entire heating energy system. The calculation formula for the minimization of system operation cost is In the formula, C t is the total system operation cost, C f is the fuel cost, C m is the equipment maintenance cost, C p is the energy purchase cost from the external power grid or gas network. T is the total number of time periods. i ∈ CHP is the set of numbers of combined heat and power equipment. a i is the fuel cost coefficient per unit power generation of combined heat and power equipment i, is the power generation power of combined heat and power equipment i at time t. b i is the fuel cost coefficient per unit heat supply of combined heat and power equipment i, is the power generation power of combined heat and power equipment i at time t. j ∈ GB is the set of numbers of gas filtration equipment. c j is the fuel cost coefficient per unit heat supply of gas boiler equipment j, is the heat supply power of gas boiler equipment j at time t. k ∈ equip is the set of numbers of all equipment. d k is the unit power maintenance cost coefficient of equipment k. P k,t is the operating power of equipment k at time t. λ grid,t 、P grid,t 、λ gas,t and G gas,t are the electricity price for purchasing electricity from the power grid, the electric power purchased from the power grid, the gas price for purchasing gas from the gas network, and the gas volume purchased from the gas network at time t respectively. The calculation formula for the maximization of energy utilization rate is In the formula, η t is the total energy utilization rate of the system, and E input,tThe power load demand of the system, the thermal load demand of the system, the total energy input to the system, and η at time t, respectively conv is the energy conversion efficiency. The calculation formula for carbon emission minimization is In the formula, CO2 is the total carbon emission of the system, and γ CHP and γ GB are the carbon emission coefficients per unit power generation of the combined heat and power equipment and the carbon emission coefficients per unit heat supply of the gas boiler equipment, respectively;

[0078] S32. Establish the conditions of energy balance constraint, equipment operation constraint, and heat network transmission loss constraint, so that the system can stably meet the load demand of users, prevent equipment damage due to over - operation or power mutation, and conform to the actual system operation situation;

[0079] S33. Select the improved non - dominated sorting genetic algorithm combined with the multi - objective optimization model and constraint conditions. Through coding and initialization, convert the decision variables into chromosomes, and the fitness calculation can evaluate the advantages and disadvantages of each individual under multiple objectives;

[0080] S34. Output the optimal scheduling plan according to the multi - objective optimization model and energy constraint conditions.

[0081] In this embodiment, in step S32, the energy balance constraint conditions are divided into power balance constraint and heat balance constraint. The calculation formula for power balance constraint is In the formula, P grid,t 、P solar,t 、 and P ET,t are the power generation power of the combined heat and power equipment, the electric power purchased from the power grid, the power generation power of solar power generation, the power load demand of the system, and the power consumption of the electric boiler at time t, respectively. The calculation formula for heat balance constraint is In the formula, and are the heat supply power of the combined heat and power equipment, the heat supply power of the gas boiler, the heat release power of the heat storage device, the thermal load demand of the system, and the heat charging power of the heat storage device at time t, respectively. The equipment operation constraint conditions are divided into power limit constraint and ramp rate limit constraint. The calculation formula for power limit constraint is In the formula, and are the minimum power generation power and the maximum power generation power of the combined heat and power equipment i, respectively, and are the minimum heat supply power and the maximum heat supply power of the gas boiler equipment j, respectively. The calculation formula for ramp rate limit constraint is In the formula, R down and Rup They are the maximum decreasing rate and the maximum increasing rate of the device's power generation respectively. The calculation formula for the heat network transmission loss constraint condition is In the formula, and are the heat actually delivered to the user end of the heat network and the total heat generated by the heat source at time t respectively, and α loss is the heat network transmission loss rate.

[0082] In this embodiment, in step S33, the fitness calculation can evaluate the advantages and disadvantages of each individual under multiple objectives, including the following steps:

[0083] S331. Perform non-dominated sorting on all individuals in the population, divide the individuals into different non-dominated levels, and calculate their crowding distance. The calculation formula is In the formula, X i+1 and X i-1 are the next individual and the previous individual of the non-dominated individual X i respectively, and d i is the crowding distance;

[0084] S332. Randomly select pairs of parent individuals from the parent individuals obtained by the selection operation to generate offspring individuals. The calculation formula is In the formula, X a and X b are each pair of parent individuals, and X c and X d are each pair of offspring individuals

[0085] S333. For the offspring individuals obtained after crossover, determine the individuals that need to mutate with the mutation probability, and calculate the mutated components through polynomial mutation. The calculation formula is In the formula, x′ ij is the new value of the j-th component of the individual X i after the polynomial mutation operation, x ij is the original value of the j-th component of the individual X i that needs to mutate, r is a random number uniformly distributed in the range of [0 - 1], and Δ1 and Δ2 are intermediate variables for calculating the mutation amount, and where, and are the upper and lower limits of the value range of x ij respectively, and η m is the mutation distribution index;

[0086] S334. Combine the parent population and the offspring population after crossover and mutation into a combined population. The calculation formula is R t = P t ∪ Q t, where R t is the merged population, P t is the parent population, Q t is the offspring population. Perform non - dominated sorting on the merged population, and successively select individuals in the non - dominated levels to join the new population P t+1 , until the number of individuals in the new population reaches the population size N;

[0087] S335. Check whether the current iteration number t reaches the preset maximum iteration number T max and calculate the difference between the non - dominated solution sets obtained from two adjacent iterations. When t≥T max and the difference is less than the preset threshold ∈, the algorithm terminates, and the non - dominated solutions in the current population are output as the optimization results.

[0088] Specifically, by using the probability distribution estimation method based on big data, the uncertainty factors of meteorological data and heat load prediction are quantitatively analyzed, and chance constraints are constructed and incorporated into the optimal scheduling model, so that the model can generate a more robust scheduling plan under the condition of meeting a certain confidence level, greatly reducing the system operation risk.

[0089] In this embodiment, in step S4, constructing chance constraints and reducing system operation risk include the following steps:

[0090] S41. Collect data of historical temperature, humidity, and wind speed, and calculate the temperature correction coefficient. The calculation formula is where α temp is the temperature correction coefficient, T actual is the actual temperature, T base is the loss temperature. Statistically analyze the historical heat load curve and extract the peak - valley characteristics. The calculation formula is ΔQ pv =Q max -Q min , where ΔQ pv is the peak - valley heat load difference, Q max and Q min are the maximum and minimum values of the heat load respectively, and calculate the prediction error rate. The calculation formula is where ∈ t is the prediction error rate, Q f,t and Q a,t are the predicted value and the actual value of the heat load at time t respectively;

[0091] S42. Analyze the distribution characteristics of heat load data, meteorological data, etc. The calculation formula is where f(x) is the probability density value at the data point x, n is the number of samples, h is the bandwidth parameter, x iis the i-th sample data point, K(·) is the Gaussian kernel function, a joint probability graph model of meteorology and load is constructed, and the uncertainty is quantified through the posterior probability P(Q load |T,W), where T is temperature and W is humidity;

[0092] D43. Describe the uncertain parameters using the gradient fuzzy number algorithm, and the calculation formula is where μ(x) is the membership function of the independent variable x for the given trapezoidal fuzzy set, and a, b, c, and d are all parameters of the trapezoidal fuzzy number;

[0093] S44. Establish chance-constrained transformation to ensure that the power supply for heating energy meets the load demand. The chance-constrained transformation includes power balance constraint and equivalent crisp constraint, and the calculation formula is where Pr(·) is the probability operation, ∑ i P i,t is the sum of the power generation of all power generation equipment at time t, β is the confidence level, δ is the coefficient related to the confidence level β, and δ = Φ -1 (1-β), where Φ - (·) is the inverse cumulative distribution function of the standard normal distribution, and σ load is the standard deviation of the power load demand;

[0094] S45. Simulate uncertain factors through random sampling and construct a two-stage robust optimization model. Considering the situation of uncertainty, minimize the total cost, and the calculation formula is where is the load value at time t and the s-th load scenario, is the mean μ t , variance of the normal distribution, min x is the minimization operation of the variable x, is the expectation operation, C(x,ζ) is the cost function, ρ is the risk aversion coefficient, and the value range is [0-1], CVaR α (·) is the conditional value at risk, α is the confidence level, and the value range is [0-1], is the scenario probability π s and the adjustment amount at time t under all scenarios s The sum of the products, is the scenario probability π s and the load value at time t under all scenarios s The sum of the products;

[0095] S46. In the scenario of dynamic update of uncertain parameters, optimize the model parameters in real time according to the continuously obtained new data, and dynamically adjust the threshold according to the fluctuation of the load data. The calculation formula is wherein, P(θ|D 1:t ) is the posterior probability distribution of the model parameter θ obtained from the historical data set D 1:t from time 1 to time t, P(D t |θ) is the likelihood function, β t is the dynamic threshold at time t, β0 is the initial threshold, γ is the adjustment factor, and Var(Q load,t ) is the variance of the load Q load,t at time t;

[0096] S47. During the time period, minimize the sum of the fuel cost and the reserve cost to ensure that the total power generation of the power generation equipment can meet the load demand. The calculation formula is min∑ t (C f,t +C r,t ), where C f,t is the fuel cost at time t, and C r,t is the reserve cost at time t.

[0097] Specifically, by leveraging big data technology to mine potential influencing factors, the driving factors of the heat load change can be understood more comprehensively, thereby realizing more refined energy scheduling. After considering these potential factors, the system can more reasonably arrange the power generation and heat supply powers of the heat source equipment, optimize the energy distribution plan, improve the energy utilization efficiency, and reduce the system operation cost.

[0098] In this embodiment, in step S5, the generated adjustment strategy is divided into heat source output adjustment and heat network dynamic regulation. The priority order of the heat source output adjustment is to start the standby boiler, adjust the CHP unit, and call the energy storage. The calculation formula is where P GB,t , P CHP,t and are respectively the power generation power of the standby boiler, the power generation power of the CHP unit, and the discharge amount of the energy storage system at time t, P GB,max is the maximum power generation power of the standby boiler, ΔQ and ΔQ GB,max are respectively the heat load change amount and the maximum heat supply capacity of the standby boiler, is the smaller value of taking 1 and to prevent exceeding its maximum power generation power, R up is the rising power adjustment rate of the CHP unit, Δt is the time interval, Q ST,state is the current power state of the energy storage system, and η dis is the discharge efficiency of the energy storage system, is the smaller value of taking Q ST,state and The smaller value among them ensures that the discharge amount of the energy storage system does not exceed its remaining power. The dynamic regulation of the heat network is controlled by adjusting the valve opening, and the calculation formula is In the formula, α v,t is the opening of the heat network valve at time t, K p is the proportionality coefficient, with a value of 0.5, ΔT t is the temperature deviation at time t, K i is the integral coefficient, with a value of 0.1, ∫ΔT t dt is the integral value of the temperature deviation over time, K d is the differential coefficient, with a value of 0.2, is the rate of change of the temperature deviation.

[0099] Specifically, by making rapid adjustments according to real-time data and dynamic changes, the heating energy system has stronger adaptability and flexibility. Whether it is dealing with sudden changes in weather, fluctuations in user heat load demand, or emergencies such as equipment failures, the system can respond in a timely manner, adjust the scheduling strategy, ensure the quality of heating services, improve the heating satisfaction of residents, and at the same time provide strong support for the sustainable management of urban energy.

[0100] In this embodiment, in step S6, when a deviation occurs, starting the dynamic adjustment mechanism includes the following steps:

[0101] S61. Set the trigger threshold for starting the dynamic adjustment mechanism, and the trigger condition is In the formula, L is the emergency adjustment, N is the normal adjustment, ΔP t is the power deviation, ΔT t is the temperature deviation, and calculate the deviation in real time. The calculation formula is In the formula, P a,t and P s,t are the actual power and planned power at time t respectively, T u,t and T s,t are the supply temperature and actual temperature at time t;

[0102] S62. By inputting the feature vector at the current moment and the updated model parameters, obtain the load prediction value for the next moment. The calculation formula is In the formula, is the load prediction value for the next moment, M(·) is the long short-term memory network model, X t is the feature vector at time t, θ new is the updated LSTM model parameter;

[0103] S63. Evaluate the heating optimization regulation performance, which reflects the degree of correction of the adjustment strategy to the heat load at different times. The calculation formula is where ξ is the adjustment efficiency, ΔQ c,t is the changed heat load after adjustment at time t, ΔQ i,t is the initial changed heat load at time t, ΔC is the changed cost, C a is the cost after adjustment, C o is the original cost;

[0104] S64. Establish a feedback mechanism to correct the conversion efficiency parameter. The calculation formula is where η conv,new and η conv,old are the conversion efficiency parameters after and before correction respectively, MSE load is the mean square error of load prediction, Q mean is the load mean value;

[0105] S65. Use the transfer learning method, combine the basic model and real-time data to obtain the final model parameters. The calculation formula is θ final =TransferLearning(f(x),x s ), where θ final is the final model parameter, TransferLearning(·) is the transfer learning algorithm, f(x) is the basic model, x s is the real-time data.

[0106] In this embodiment, in step S11, the sensors include but are not limited to temperature sensors, pressure sensors, flow sensors, power sensors, and the data acquisition frequency is dynamically adjusted according to different data types and system requirements to ensure that the collected data can accurately reflect the real-time operation state of the system.

[0107] Specifically, by real-time monitoring the actual operation state of the system, once it is detected that there is a deviation between the actual operation situation and the scheduling scheme, the system will promptly start the dynamic adjustment mechanism, based on the latest monitoring data, re-perform load prediction and optimal scheduling calculation, and online adjust the scheduling scheme to ensure that the system is always in the optimal operation state. This real-time monitoring and dynamic adjustment mechanism can quickly respond to changes in the system, improve the reliability and stability of the system. At the same time, by real-time adjusting the opening degrees of the regulating valves in the heat network, the economy and stability of the heat network operation are ensured, and the operation efficiency of the entire heating energy system is further improved.

[0108] The above embodiments are only used to illustrate the present invention, rather than to limit the present invention. Although the present invention has been described in detail with reference to the embodiments, those of ordinary skill in the art should understand that various combinations, modifications or equivalent replacements of the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and should all be covered within the scope of the claims of the present invention.

Claims

1. A heating energy system optimization scheduling method based on big data processing, characterized in that: The following steps are involved: S1. Collect various data in the heating energy system, pre-process the data, and build a large database of the heating energy system; S2. Based on the preprocessed data, the heat load prediction model is trained in combination with historical data, and the potential factors affecting the heat load change are explored using big data analysis technology; S3, taking the load forecast results as known inputs and considering the heat loss during the heat network transmission process, an optimal scheduling model for the heating energy system is established; S4. Use probability distribution estimation methods based on big data to quantitatively analyze uncertainty factors, build opportunity constraints, and reduce system operation risks; S5. Use intelligent optimization algorithm to solve the optimization scheduling model and generate detailed intelligent optimization strategy according to the solution results; S6. Monitor the actual operating status of the system in real time, and when deviations occur, promptly activate the dynamic adjustment mechanism to ensure that the system is always in the optimal operating state.

2. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 1 is characterized in that: In step S1, building a large database of the heating energy system includes the following steps: S11. Use various sensors distributed in the heating energy system to collect real-time meteorological data, heat source equipment operation data, heat network operation data and user heat load data; S12, perform anomaly detection on the collected data and correct the abnormal data. The calculation formula is: In the formula, μ is the sliding window mean, σ is the sliding window standard deviation, and missing values ​​are interpolated to ensure data integrity. The calculation formula is: In the formula, x t is the value of the variable at time t, is the variable value at time t+i, and NN is the predicted value obtained by the neural network algorithm; S13. Unify the data format, add timestamps and labels to the data, and establish a time series database; S14. Check the data integrity and verify the data consistency; S15. Add label indexes to the data and perform distributed storage of the data.

3. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 1 is characterized in that: In step S2, mining potential factors affecting heat load changes includes the following steps: S21. Statistical analysis of the heat load variation patterns in different time periods, analysis of the impact of residents’ work and rest schedules on heat load, and calculation of the average heat load in each time period. The calculation formula is In the formula, is the average heat load, is the heat load at the tth moment in the i-th time period, and n is the number of data in this time period; S22, taking meteorological data, date type, time information and mined potential influencing factors as input features of the model; S23, the model input features are divided into training set, validation set and test set in a ratio of 7:2:1, and random forest regression is used to train on the randomly selected feature subsets. The calculation formula is In the formula, is the predicted value after training, m is the number of decision trees, y j is the predicted value of the jth decision tree; S24. Use the validation set and test set to evaluate the trained model. The calculation formula is: In the formula, MSE is the evaluation result, y i is the true value, is the predicted value, and n is the number of samples.

4. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 1 is characterized in that: In step S3, establishing an optimal scheduling model for the heating energy system includes the following steps: S31. Taking minimizing system operating costs, maximizing energy utilization, and minimizing carbon emissions as comprehensive goals, a multi-objective optimization model is constructed to lay the foundation for optimizing the scheduling of the entire heating energy system. The system operating cost minimization calculation formula is: In the formula, C t is the total operating cost of the system, C f is the fuel cost, C m is the equipment maintenance cost, C p is the cost of purchasing energy from the external power grid or gas grid, T is the total number of time periods, i∈CHP is the set of CHP equipment numbers, a i is the fuel cost coefficient of unit power generation of CHP equipment i, is the power generation capacity of cogeneration equipment i at time t, b i is the fuel cost coefficient for unit heat supply of CHP plant i, is the power generation of cogeneration equipment i at time t, j∈GB is the serial number collection of gas filtration equipment, c j is the fuel cost coefficient of unit heat supply of gas boiler equipment j, is the heating power of gas boiler equipment j at time t, k∈equip is the serial number collection of all equipment, d k is the unit power maintenance cost coefficient of equipment k, P k,t is the operating power of device k at time t, λ grid,t , P grid,t , gas,t and G gas,t are the electricity price purchased from the power grid at time t, the electric power purchased from the power grid, the gas price purchased from the gas grid, and the gas volume purchased from the gas grid. The calculation formula for maximizing energy utilization is: Where η t is the total energy utilization rate of the system, and E input,t are the power load demand of the system at time t, the thermal load demand of the system, and the total energy input to the system, η conv is the energy conversion efficiency, and the calculation formula for minimizing carbon emissions is: In the formula, CO2 is the total carbon emission of the system, γ CHP and γ GB They are the carbon emission coefficient per unit of electricity generated by cogeneration equipment and the carbon emission coefficient per unit of heat provided by gas boiler equipment; S32. Establish conditions for energy balance constraints, equipment operation constraints, and heat network transmission loss constraints, so that the system can stably meet the user's load demand, prevent equipment from being damaged due to excessive operation or power mutation, and conform to the actual system operation conditions; S33, select the improved non-dominated sorting genetic algorithm combined with the multi-objective optimization model and constraints, convert the decision variables into chromosomes through encoding and initialization, and the fitness calculation can evaluate the pros and cons of each individual under multiple objectives; S34. Output the optimal scheduling plan based on the multi-objective optimization model and energy constraints.

5. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 4 is characterized in that: In step S32, the energy balance constraint condition is divided into power balance constraint and thermal balance constraint. The power balance constraint calculation formula is: In the formula, and P ET,t are the power generation power of the cogeneration equipment at time t, the power purchased from the power grid, the power of solar power generation, the power load demand of the system and the power consumption of the electric boiler. The thermal balance constraint calculation formula is: In the formula, and They are the heating power of the cogeneration equipment, the heating power of the gas boiler, the heat release power of the heat storage device, the thermal load demand of the system and the charging power of the heat storage device at time t. The equipment operation constraints are divided into power limit constraints and ramp rate limit constraints. The power limit constraint calculation formula is: In the formula, and are the minimum power generation and maximum power generation of cogeneration equipment i, respectively. and are the minimum heating power and the maximum heating power of the gas boiler equipment j respectively. The calculation formula of the ramp rate limit constraint is: In the formula, R down and R up are the maximum decreasing rate and the maximum increasing rate of the equipment power generation respectively. The calculation formula of the heat network transmission loss constraint condition is: In the formula, and are the heat actually delivered to the user end by the heat network at time t and the total heat generated by the heat source, α loss is the heat network transmission loss rate.

6. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 4 is characterized in that: In step S33, the fitness calculation can evaluate the pros and cons of each individual under multiple objectives and includes the following steps: S331. Perform non-dominated sorting on all individuals in the population, divide the individuals into different non-dominated layers, and calculate their crowding distance. The calculation formula is: Where, X i+1 and X i-1 are the non-dominated individuals X i The next individual and the previous individual, d i is the crowding distance; S332. Randomly select pairs of parent individuals from the parent individuals obtained by the selection operation to generate offspring individuals. The calculation formula is: Where, X a and X b For each pair of parent individuals, X c and X d For each pair of offspring S333. For the offspring individuals obtained after crossover, the individuals that need to be mutated are determined by mutation probability, and the components after mutation are calculated by polynomial mutation. The calculation formula is: In the formula, x′ ij is the individual X after polynomial mutation operation i The new value of the jth component in x ij X is the individual that needs to be mutated i The original value of the jth component in , r is a random number uniformly distributed in the interval [0-1], Δ1 and Δ2 are intermediate variables for calculating the variation, and in, and x ij The upper and lower limits of the value range, η m is the variation distribution index; S334, merge the parent population and the offspring population after crossover and mutation into a merged population, and the calculation formula is R t =P t ∪Q t , where R t To merge the population, P t is the parent population, Q t For the offspring population, the merged population is sorted by non-dominated order, and individuals in the non-dominated layer are selected in turn to join the new population P t+1 , until the number of individuals in the new population reaches the population size N; S335: Check whether the current number of iterations t reaches the preset maximum number of iterations T max And calculate the difference between the non-dominated solution sets obtained in two consecutive iterations. When t≥T max When the difference is less than the preset threshold ∈, the algorithm terminates and outputs the non-dominated solution in the current population as the optimization result.

7. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 1 is characterized in that: In step S4, constructing opportunity constraints to reduce system operation risks includes the following steps: S41. Collect historical temperature, humidity, and wind speed data and calculate the temperature correction coefficient. The calculation formula is: In the formula, α temp is the temperature correction factor, T actual is the actual temperature, T base is the loss temperature, the historical heat load curve is statistically analyzed, the peak and valley characteristics are extracted, and the calculation formula is ΔQ pv =Q max -Q min , where ΔQ pv is the peak-valley heat load difference, Q max and Q min are the maximum and minimum values ​​of the heat load respectively, and the prediction error rate is calculated. The calculation formula is: In the formula, ∈ t is the prediction error rate, Q f,t and Q a,t are the predicted value and actual value of heat load at time t respectively; S42. Analyze the distribution characteristics of heat load data, meteorological data, etc. The calculation formula is: Where f(x) is the probability density value at data point x, n is the number of samples, h is the bandwidth parameter, and x i is the i-th sample data point, K(·) is the Gaussian kernel function, and the joint probability graph model of meteorology and load is constructed. load |T,W) to quantify uncertainty, where T is temperature and W is humidity; D43, using the gradient fuzzy number algorithm to describe uncertain parameters, the calculation formula is: In the formula, μ(x) is the membership function of the independent variable x to the given trapezoidal fuzzy set, and a, b, c, and d are all parameters of the trapezoidal fuzzy number; S44. Establish opportunity constraint conversion to ensure that the power supply and heating energy meet the load demand. The opportunity constraint conversion includes power balance constraint and equivalent clear constraint. The calculation formula is: Where Pr(·) is the probability operation, ∑ i P i,t is the sum of the power generated by all power generation equipment at time t, β is the confidence level, δ is the coefficient related to the confidence level β, and δ = Φ -1 (1-β), where Φ - (·) is the inverse cumulative distribution function of the standard normal distribution, σ load is the standard deviation of power load demand; S45. We simulate uncertainties through random sampling and build a two-stage model for robust optimization. We minimize the total cost under uncertainty. The calculation formula is: In the formula, is the load value at time t and the sth load scenario, is the mean μ t ,variance Normal distribution, min x is the minimization operation of variable x, is the expectation operation, C(x,ζ) is the cost function, ρ is the risk aversion coefficient, the value range is [0-1], CVaR α (·) is the conditional risk value, α is the confidence level, ranging from [0-1], is the scene probability π s The adjustment amount at time t for all scenarios s The sum of products, is the scene probability π s And the load value at time t in all scenarios s The sum of the scores; S46. In the scenario of dynamic update of uncertain parameters, the model parameters are optimized in real time according to the new data continuously obtained, and the threshold is dynamically adjusted according to the fluctuation of load data. The calculation formula is: In the formula, P(θ|D 1:t ) is the historical data set D from time 1 to time t 1:t The posterior probability distribution of the obtained model parameter θ, P(D t |θ) is the likelihood function, β t is the dynamic threshold at time t, β0 is the initial threshold, γ is the adjustment factor, Var(Q load,t ) is the load Q at time t load,t The variance of S47. Minimize the sum of fuel cost and standby cost within the time period to ensure that the total power generated by the power generation equipment can meet the load demand. The calculation formula is min∑ t (C f,t +C r,t ), where C f,t is the fuel cost at time t, C r,t The backup cost at time t.

8. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 1 is characterized in that: In step S5, the generated adjustment strategy is divided into heat source output adjustment and heat network dynamic adjustment. The priority of heat source output adjustment is to start the standby boiler, adjust the CHP unit and call the energy storage. The calculation formula is: Where P GB,t , P CHP,t and are the power generation power of the standby boiler, the power generation power of the CHP unit and the discharge capacity of the energy storage system at time t, respectively. GB,max is the maximum power generation of the standby boiler, ΔQ and ΔQ GB,max are the heat load change and the maximum heating capacity of the standby boiler, To take 1 and The smaller value in the above formula prevents the maximum power generation from exceeding the R up is the rising power regulation rate of the CHP unit, Δt is the time interval, Q ST,state is the current state of charge of the energy storage system, η dis is the discharge efficiency of the energy storage system, To obtain Q ST,state and The smaller value in ensures that the discharge of the energy storage system does not exceed its remaining power. The dynamic regulation of the heating network is controlled by adjusting the valve opening. The calculation formula is: In the formula, α v,t is the opening degree of the heating network valve at time t, K p is the proportionality coefficient, the value is 0.5, ΔT t is the temperature deviation at time t, K i is the integral coefficient, with a value of 0.1, ∫ΔT t dt is the integral value of temperature deviation over time, K d is the differential coefficient, with a value of 0.

2. is the rate of change of temperature deviation.

9. The method for optimizing and scheduling a heating energy system based on big data processing according to claim 1 is characterized in that: In step S6, when a deviation occurs, starting the dynamic adjustment mechanism includes the following steps: S61. Setting a trigger threshold for starting a dynamic adjustment mechanism. The trigger condition is: Where L is the emergency adjustment, N is the routine adjustment, ΔP t is the power deviation, ΔT t is the temperature deviation, and the deviation is calculated in real time. The calculation formula is: Where P a,t and P s,t are the actual power and planned power at time t, T u,t and T s,t are the supply temperature and actual temperature at time t; S62. By inputting the characteristic vector at the current moment and the updated model parameters, the load forecast value at the next moment is obtained. The calculation formula is: In the formula, is the load forecast value at the next moment, M(·) is the long short-term memory network model, X t is the eigenvector at time t, θ new are the updated LSTM model parameters; S63. Evaluate the heating optimization regulation performance to reflect the degree of correction of the heat load by the adjustment strategy at different times. The calculation formula is: Where ξ is the adjustment efficiency, ΔQ c,t is the change in heat load after adjustment at time t, ΔQ i,t is the initial heat load change at time t, ΔC is the cost change, C a is the adjusted cost, C o is the original cost; S64. Establish a feedback mechanism to correct the conversion efficiency parameters. The calculation formula is: Where η conv,new and η conv,old are the conversion efficiency parameters before and after correction, MSE load is the mean square error of load forecast, Q mean is the load mean; S65. Use the transfer learning method to combine the basic model and real-time data to obtain the final model parameters. The calculation formula is θ final =TransferLearning(f(x),x s ), where θ final is the final model parameter, TransferLearning(·) is the transfer learning algorithm, f(x) is the basic model, x s For real-time data.

10. The method for optimizing and dispatching a heating energy system based on big data processing according to claim 2 is characterized in that: In step S11, the sensors include but are not limited to temperature sensors, pressure sensors, flow sensors, and power sensors, and the data collection frequency is dynamically adjusted according to different data types and system requirements to ensure that the collected data can accurately reflect the real-time operating status of the system.

Images