Heating furnace control method and system based on expert control and storage medium

Abstract

The invention discloses a heating furnace control method and system based on expert control and a storage medium, and the method comprises the following steps: constructing an expert knowledge employing a generative rule to represent combustion control knowledge, and the rule form is IFlt; a precondition gt; tHNt, THNt, THNt; performing conclusion operation gt; the preconditions comprise a furnace temperature error range, a furnace pressure state, a flue gas temperature and a reversing period parameter, and the conclusion operation comprises gas flow adjustment, air-fuel ratio correction, and setting of a flue valve opening degree or a burner reversing period; dynamic parameter prediction and optimization: calculating a predicted temperature value under the current working condition based on a furnace temperature prediction model; performing multi-target collaborative reasoning: searching a matching rule set in a heuristic manner, and combining a Hash algorithm to quickly position a control conclusion; the optimized air-fuel ratio, reversing period and furnace pressure parameters are issued to a PLC, and an execution mechanism is driven to adjust the frequency of a gas valve, the frequency of an air valve and the frequency of a smoke exhaust fan; and self-learning and knowledge base updating: dynamically correcting the rule weight according to the historical control effect.

Description

Technical Field

[0001] The present invention relates to the technical field of heating furnace control, and in particular to a heating furnace control method, system and storage medium based on expert control. Background Art

[0002] Heating furnaces are the most widely used heating equipment in casting and heat treatment, and they are also the largest energy consumers. Regenerative heating furnaces are used to heat materials in stages. Their temperature parameters significantly influence material preparation and production, and the accuracy and stability of temperature control directly impact forging performance and energy efficiency. Furthermore, combustion control in heating furnaces is susceptible to random interference and multiple influencing factors, resulting in a nonlinear system with large inertia, pure hysteresis, and distributed parameters. These control variables primarily include regenerative burner reversing control, furnace temperature control, furnace pressure control, and exhaust gas temperature control. These controlled variables are strongly coupled, meaning that burner reversing can significantly impact furnace pressure. Adjusting the exhaust gas flow to control furnace pressure also affects exhaust temperature, while adjusting the exhaust gas flow to control exhaust temperature also affects furnace pressure. This makes control very difficult, making it difficult to ensure stable billet heating quality, leading to severe oxidation and burning damage during billet heating and significant energy waste.

[0003] Therefore, a heating furnace optimization controller is set up, and an appropriate control method is used to set and adjust the air and gas flow rates, switching cycles, etc., in order to stabilize the furnace and flue gas temperatures, stabilize the furnace pressure, achieve stable billet heating quality, and save energy. However, because its combustion process is more susceptible to random interference than traditional combustion methods, with many influencing factors, large inertia, and hysteresis nonlinear characteristics, traditional heating furnace combustion control systems rely on precise measurement of various process parameters. At the same time, due to the large fluctuations in low calorific value gas pressure and calorific value, the impact of switching on the furnace pressure at the moment, and the uncontrollable furnace pressure, it will cause problems for the operation of the heating furnace. In addition, the traditional PID controller is not ideal in controlling the heating furnace temperature, resulting in unsatisfactory combustion performance of the heating furnace. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a heating furnace control method, system, and storage medium based on expert control. These methods stabilize furnace and flue gas temperatures, stabilize furnace pressure, achieve stable billet heating quality, and save energy.

[0005] According to a first aspect of the present invention, a heating furnace control method based on expert control includes the following steps:

[0006] Step S100: Constructing an expert knowledge base: Using production rules to represent combustion control knowledge, the rule format is IF <precondition> THEN <conclusion operation>, where the preconditions include furnace temperature error range, furnace pressure state, flue gas temperature, and reversing cycle parameters, and the conclusion operation includes gas flow adjustment, air-fuel ratio correction, flue valve opening, or burner reversing cycle setting;

[0007] Step S200, dynamic parameter prediction and optimization: Calculate the predicted temperature value under the current working condition based on the furnace temperature prediction model, the model is:

[0008] T m (k+1)=f(T(k),T(k-1),gas(k),gas(k-1),gas(k-2))

[0009] Among them, T m To predict the furnace temperature, gas is the gas flow rate, f is the nonlinear function; k is the time;

[0010] Step S300, multi-objective collaborative reasoning: quickly locate the control conclusion by searching the matching rule set using a heuristic search and combining the hash algorithm. The heuristic function is:

[0011] e(x)=G(x)―M(x)

[0012] Where G(x) is the weighted thermocouple measured temperature, and M(x) is the model predicted value;

[0013] Step S400: Execute control instructions: send the optimized air-fuel ratio, reversing cycle and furnace pressure parameters to the PLC, and drive the actuator to adjust the gas valve, air valve and exhaust fan frequency;

[0014] Step S500, self-learning and knowledge base update: dynamically modify rule weights based on historical control effects, add or delete rules, and optimize knowledge base priority sorting.

[0015] According to some embodiments of the present invention, the production rules are stored by priority, and the rule set priority is divided according to the furnace temperature error range:

[0016] When E T <―M, the air-fuel ratio adjustment rule is executed first;

[0017] When|E T When |≤M, the furnace pressure stabilization rule is executed first;

[0018] When E T >M, the gas flow increase rule is executed first;

[0019] Among them, E Tis the error value between the model predicted temperature and the actual detected temperature, and M is the preset threshold.

[0020] According to some embodiments of the present invention, the heuristic search specifically includes:

[0021] Filter the initial rule set through SQL statements;

[0022] A hash table is used to establish the mapping relationship between the premise and the conclusion, and the conclusion is directly located through the LookUp() function;

[0023] If the match fails, the self-learning module is triggered to generate new rules and store them in the knowledge base.

[0024] According to some embodiments of the present invention, the self-learning module updates the knowledge base in the following manner:

[0025] Collect statistics on historical control effects and calculate the success rate of rule execution;

[0026] If the success rate is lower than the threshold, the rule priority is lowered or marked as invalid;

[0027] If the newly added working condition data conflicts with the existing rules, new rules are generated and assigned initial weights.

[0028] According to some embodiments of the present invention, the commutation period includes:

[0029] Dynamically adjust the burner reversing batches according to the exhaust gas temperature curve, adopt a dispersed reversing method, and only reversing part of the burners each time;

[0030] When the reversing command is triggered, the exhaust fan frequency is adjusted synchronously to offset the furnace pressure shock.

[0031] According to some embodiments of the present invention, the air-fuel ratio correction includes:

[0032] Dynamically adjust the air flow rate based on the deviation between the measured and predicted flue gas oxygen content;

[0033] When the deviation exceeds the preset value, the air-fuel ratio emergency correction rule is triggered to limit the increase in gas flow.

[0034] According to some embodiments of the present invention, the further comprising:

[0035] Integrate fuzzy logic algorithm into the inference engine to fuzzify the rule premise conditions;

[0036] When there is a conflict in rule matching, the membership function is used to weight and generate the final control instruction.

[0037] According to some embodiments of the present invention, the expert knowledge base supports offline expansion, including:

[0038] Add new rules through manual input and assign initial priorities;

[0039] Import external data sources to generate generalization rules.

[0040] According to an embodiment of the second aspect of the present invention, a heating furnace control system includes a memory and a processor, wherein the memory stores a determination program for heating furnace control based on expert control, and the processor runs the determination program for heating furnace control based on expert control to enable the heating furnace control system to execute the above-mentioned heating furnace control method based on expert control.

[0041] According to a third aspect of an embodiment of the present invention, a computer-readable storage medium includes: a determination program for heating furnace control based on expert control is stored on the computer-readable storage medium, and when the determination program for heating furnace control based on expert control is executed by a processor, the above-mentioned heating furnace control method based on expert control is implemented.

[0042] The heating furnace control method, system and storage medium based on expert control according to the embodiment of the present invention have at least the following features:

[0043] Beneficial effects:

[0044] 1. By establishing a furnace temperature prediction model, the control parameters can be adjusted in advance to reduce the fluctuation range of furnace temperature and improve the stability of furnace temperature.

[0045] 2. Combining heuristic search and hash algorithm, the control specified generation time is short, the response speed is greatly improved compared with the traditional PID, and the hysteresis is reduced.

[0046] 3. Adjust the air flow in real time according to the deviation of flue gas oxygen content, reduce the air-fuel ratio error, improve thermal efficiency, and achieve the effect of energy saving and thermal efficiency optimization.

[0047] 4. The use of decentralized reversing control can reduce the impact of furnace pressure during the reversing process, which is beneficial to the stable control of furnace pressure and reduces gas consumption.

[0048] 5. A self-learning and knowledge base has been established to enable the addition of heating furnace expert knowledge base rules and provide more reasonable and optimized parameters.

[0049] 6. Compared with other control schemes that optimize a single parameter, the present invention optimizes and sets parameters for the furnace temperature, furnace pressure, switching cycle, flue gas temperature, induced draft fan frequency, etc. as a whole, alleviating the mutual coupling between the controlled quantities and making the operation of the heating furnace more stable.

[0050] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0052] Figure 1 This is a flowchart of a heating furnace control method based on expert control according to an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the overall structure of a heating furnace control system according to an embodiment of the present invention;

[0054] Figure 3 A schematic diagram of the structure of a heating furnace expert system according to an embodiment of the present invention;

[0055] Figure 4 This is a general diagram of combustion knowledge of a heating furnace according to an embodiment of the present invention;

[0056] Figure 5 A schematic diagram of a heating furnace combustion reasoning mechanism according to an embodiment of the present invention;

[0057] Figure 6 This is a forward reasoning block diagram of heating furnace combustion according to an embodiment of the present invention;

[0058] Figure 7 This is a structural diagram of a heating furnace control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0060] In the description of the present invention, it should be understood that descriptions involving orientation, such as the orientation or positional relationship indicated by up, down, etc., are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0061] In the description of the present invention, "a plurality" refers to more than two. The use of "first" or "second" is solely for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of the indicated technical features, or implicitly indicating the order of the indicated technical features.

[0062] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, and connecting should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention based on the specific content of the technical solution.

[0063] Reference Figure 1 As shown, the present invention discloses a heating furnace control method based on expert control, comprising the following steps:

[0064] Constructing an expert knowledge base: Using production rules to represent combustion control knowledge, the rule format is IF <precondition> THEN <conclusion operation>. The preconditions include furnace temperature error range, furnace pressure status, flue gas temperature, and reversing cycle parameters. The conclusion operation is gas flow adjustment, air-fuel ratio correction, flue valve opening, or burner reversing cycle setting.

[0065] Dynamic parameter prediction and optimization: Calculate the predicted temperature value under the current working conditions based on the furnace temperature prediction model. The model is:

[0066] T m (k+1)=f(T(k),T(k-1),gas(k),gas(k-1),gas(k-2))

[0067] Among them, T m To predict the furnace temperature, gas is the gas flow rate, f is the nonlinear function; k is the time;

[0068] Multi-objective collaborative reasoning: By using a heuristic search matching rule set and combining it with a hash algorithm, the control conclusion can be quickly located. The heuristic function is:

[0069] e(x)=G(x)―M(x)

[0070] Where G(x) is the weighted thermocouple measured temperature, and M(x) is the model predicted value;

[0071] Execution control instructions: Send the optimized air-fuel ratio, reversing cycle and furnace pressure parameters to the PLC, driving the actuator to adjust the gas valve, air valve and exhaust fan frequency;

[0072] Self-learning and knowledge base update: Dynamically modify rule weights based on historical control effects, add or delete rules, and optimize knowledge base priority sorting.

[0073] Combine Figure 2 and Figure 3 As shown, in this embodiment, the prediction model adopts a multi-model structure, wherein the furnace temperature prediction model can be discretized by adding a zero-order holder to the furnace temperature mechanism mathematical model to obtain the furnace temperature prediction model as follows:

[0074] T m(k+1)=f(T(k),…,T(k―n a +1),gas(k),…,gas(k―n b +1))+ ε (k)

[0075] T m (k+1) represents the furnace temperature predicted by the furnace temperature prediction model; k is the time; gas(k) represents the gas flow at time k; n a and n b are the input and output orders of the system respectively; ε(k) is a white noise sequence; f is a nonlinear function, and f(·)∈L 2 (R na +nb ). Determine n through engineering experience a and n b The value of is 2,3. It can be simplified to

[0076] T m (k+1)=f(T(k),T(k-1),gas(k),gas(k-1),gas(k-2))

[0077] Other aspects, such as the relationship model between furnace pressure and air flow, will not be discussed here. Model modification and optimization, through the detection of individual data values ​​and the intelligent algorithm of the expert system, predict and manage the combustion of the heating furnace. The resulting optimized value is sent to the PLC system, instructing the underlying actuators to execute the command, thereby improving combustion efficiency and stabilizing furnace pressure.

[0078] The overall workflow is as follows: Key parameters such as furnace temperature, furnace pressure, flue gas oxygen content, and gas flow rate are first acquired. Future furnace temperatures are then predicted using the temperature prediction model Tm(k+1), providing a basis for control decisions. Based on a rule base, current operating conditions are matched and optimized control instructions are generated. The PLC drives actuators (such as gas valves and exhaust fans) to adjust their parameters. The knowledge base is dynamically updated based on execution results, improving system adaptability.

[0079] Combustion control knowledge is expressed using production rules (IF-THEN format). Rule example:

[0080] IF preheating section temperature error ≤ -40℃ ∧ furnace pressure normal ∧ flue gas temperature ≤ -100℃;

[0081] THEN gas flow +1000m 3 / h∧Flue valve opening +5%.

[0082] Under the conditions of meeting the real-time parameter combination of furnace temperature error, furnace pressure status, flue gas temperature, etc., specific control actions such as gas flow adjustment, air-fuel ratio correction, and reversing cycle setting are performed.

[0083] Combine Figure 4 As shown, by classifying the rules, according to the furnace temperature error range (such as E T <―M、|E T |≤M,E T >M) into sets of rules with different priorities. Arrange the rules with the highest priority at the beginning of each category, and the rules with the lowest priority at the end. This narrows the knowledge search scope of the expert system and improves the speed and efficiency of the inference engine.

[0084] In the furnace temperature prediction model, the current and historical furnace temperature (T(k)) and gas flow (gas(k)) parameters are input, and the predicted furnace temperature T at the next moment is output. m (k+1), thereby predicting the temperature change trend in advance and providing an advance adjustment basis for control decisions.

[0085] Through the heuristic function, the current control effect is judged by the error value, driving the rule matching and adjustment.

[0086] In heuristic search, the key is the heuristic function, which we use G(x) to represent a combustion temperature value calculated based on the current combustion status of the heating furnace.

[0087]

[0088] Where m is the number of thermocouples in the preheating section; n is the number of thermocouples in the first heating section; s is the number of thermocouples in the second heating section; p is the number of thermocouples in the soaking section; a i (x), b j (x), c l (x), d h (x) represents the weighted system, which is a constantly changing quantity, where A i (x) represents the temperature value of the i-th thermocouple in the preheating section; similarly, B j (x) represents the temperature value of the jth thermocouple in a heating section, C l (x), D h The meaning of (x) is similar. We use M(x) to represent the predicted value of the model and e(x) to represent the heuristic function. Finally, the heuristic function is as follows:

[0089] e(x)=G(x)―M(x)

[0090] After the value of e(x) is calculated, the heuristic function e(x) and the rule set classification principle E are established based on its value and combined with the rule set classification principle. TThe one-to-one correspondence between them is combined with the Open() member function of the Crecordset (Crecordset is the MFC database base class) class (with SQL (Structure Query Language) statement parameters), and then the if...else statement is used to determine and complete the heuristic search strategy.

[0091] The hash algorithm is implemented using the MFC CMap template. A CConToResultMap object is declared directly using the CMap class. The CMap class member function SetAt() then establishes a one-to-one correspondence between its conditions and conclusions. During inference, the member function LookUp() directly locates the conclusion corresponding to the rule without performing any comparisons. IsEOF() determines whether the end of the file has been reached, indicating the end of the process.

[0092] Using multi-objective collaborative reasoning, the initial rule set is screened according to the current working conditions (such as furnace temperature error, furnace pressure state) to narrow the matching range. For example, when E T <―M, prioritize matching the “heating rule set”. Convert the rule prerequisites into unique hash keys, use the CMap class to create a key-value mapping, and quickly locate through the hash algorithm. For example: Generate a hash key for the current working condition (such as “E T <=-40_Furnace pressure normal_Flue gas temperature <=-100"). The corresponding control conclusion is directly retrieved through the LookUp() function. If there is no match, the self-learning module is triggered to generate a new rule.

[0093] The optimized control parameters (such as gas flow setpoint and switching cycle) are sent to the PLC. The PLC drives the actuators to execute actions. These actions include adjusting the gas valve opening to control gas input, adjusting the air valve opening to optimize the air-fuel ratio, and controlling the exhaust fan frequency to stabilize furnace pressure.

[0094] refer to Figure 2 As shown, during this process, self-learning and knowledge base updates will be performed. Specifically, the execution success rate will be counted. If the success rate is lower than the threshold (such as 70%), the priority will be lowered or the mark will be invalid. For example, a rule has caused the furnace pressure to exceed the limit many times, and the weight is downgraded to the lowest. When no existing rules are matched, new rules are generated based on historical data. The specific process is as follows: record the current operating parameters and control requirements; simulate recommended control actions (such as gas flow +800m 3 After verification, the data is stored in the knowledge base and assigned an initial weight. This allows for the addition of rules to the heating furnace expert knowledge base, providing more reasonable and optimized parameters.

[0095] The entire control method solves the problem of multi-parameter coupling of heating furnaces through closed-loop control of expert rule matching, dynamic prediction and self-learning optimization. It has high precision, fast response and adaptability, and is suitable for complex industrial scenarios.

[0096] In some embodiments of the present invention, the production rules are stored by priority, and the rule set priority is divided according to the furnace temperature error range:

[0097] When E T <―M, the air-fuel ratio adjustment rule is executed first;

[0098] When|E T When |≤M, the furnace pressure stabilization rule is executed first;

[0099] When E T >M, the gas flow increase rule is executed first;

[0100] Among them, E T is the error value between the model predicted temperature and the actual detected temperature, and M is the preset threshold.

[0101] refer to Figure 3 As shown, E T =T 预测 -T 实际 , where T 预测 The model predicts temperature, T 实际 is the actual temperature measured by the thermocouple. M is the preset threshold used to classify the urgency of the control strategy.

[0102] In the E T When <―M (temperature is significantly low), the combustion efficiency can be improved and the furnace temperature can be quickly raised by optimizing the ratio of air to gas (such as reducing the excess air coefficient α). Example rules:

[0103] IF$E T <-20℃$∧Flue gas oxygen content>5%;

[0104] Then the air flow rate decreases by 10% and the gas flow rate increases by 8%.

[0105] When|E T When |≤M (temperature is close to the target), under the premise of temperature stability, avoid furnace pressure fluctuations caused by reversing or exhaust adjustment, and give priority to executing the furnace pressure stability rule. Example rule:

[0106] IF$|E T |≤20℃$∧Furnace pressure fluctuation>50Pa;

[0107] Then the exhaust fan frequency increases by 5Hz and the burner reversing cycle is extended by 10%.

[0108] When ET When >M (temperature is significantly higher), reduce gas supply appropriately to avoid overheating, and adjust exhaust gas to maintain furnace pressure. Example rules:

[0109] IF$ET>20℃$∧Flue gas temperature>800℃;

[0110] Then the gas flow rate decreased by 12% and the flue valve opening increased by 8%.

[0111] When the system matches multiple rule sets simultaneously, only the rule with the highest priority is executed. If multiple rules are triggered, they are executed in order of priority (air-fuel ratio adjustment > furnace pressure stabilization > gas flow adjustment), ensuring that key parameters are regulated first.

[0112] By setting the priority of the rule set, it can accurately respond to temperature deviations. T <―M, the air-fuel ratio is optimized to improve the combustion efficiency and accelerate the furnace temperature to return to the preset value. When ET<-M, the gas flow restriction rule reduces the risk of temperature overshoot. When |E T When |≤M, the uniformity of the heating elements within the furnace can be improved by reducing the fluctuation range of the furnace pressure. Furthermore, by prioritizing multiple parameters, conflicts in adjustment can be avoided, improving system robustness and reducing the probability of control failure.

[0113] Example 1:

[0114] The target temperature of a heating furnace is set to 1200°C, and the actual detected temperature is 1150°C (ET = -50°C, M = 20°C):

[0115] The system matches the ET<-M rule set, triggering the air-fuel ratio adjustment rule, reducing the excess air coefficient from 1.2 to 1.05 and increasing the gas flow by 10%;

[0116] The furnace temperature was raised to 1180°C (E T =-20℃), switch to furnace pressure stabilization rule and maintain stable exhaust fan frequency;

[0117] The final furnace temperature is stabilized at 1195℃(∣E T |=5℃), the furnace pressure fluctuation during the whole process is <±15Pa.

[0118] In summary, this embodiment prioritizes rules based on furnace temperature errors, achieving hierarchical control and multi-objective collaboration. This method improves temperature control accuracy while ensuring furnace pressure stability and energy efficiency, making it particularly suitable for highly coupled nonlinear industrial heating scenarios.

[0119] refer to Figure 5 and Figure 6 As shown, in some embodiments of the present invention, the heuristic search specifically includes:

[0120] Filter the initial rule set through SQL statements;

[0121] A hash table is used to establish the mapping relationship between the premise and the conclusion, and the conclusion is directly located through the LookUp() function;

[0122] If the match fails, the self-learning module is triggered to generate new rules and store them in the knowledge base.

[0123] In this embodiment, all control rules are stored in a relational database. Each rule contains preconditions (such as temperature error range and furnace pressure status) and conclusion operations (such as adjusting gas flow). Based on real-time collected operating parameters (such as current furnace temperature error and flue gas temperature), SQL query statements are generated to filter out potentially relevant rules. This can narrow the rule matching scope and reduce the computational complexity of subsequent processing. This avoids traversing all rules and improves efficiency. By combining conditions and filtering out irrelevant rules, matching accuracy is improved.

[0124] A hash table is used to establish a mapping relationship between the premise and the conclusion, and the premise of the rule is converted into a unique string or numerical hash key. For example, "temperature error ≤ -20℃ ∧ furnace pressure normal" is encoded as E T <= -20_Pressure_Normal. Use a hash table (such as the CMap class in C++) to store key-value pairs, where the key is the hash key and the value is the corresponding conclusion operation. Use the LookUp() function to input the hash key of the current condition and directly retrieve the conclusion. This enables fast response and control instruction generation, bypassing complex rule matching logic and avoiding redundant calculations.

[0125] When there is no corresponding key in the hash table, the self-learning module is triggered. Specifically, the current operating parameters (such as temperature error, furnace pressure, gas flow) are recorded. Based on historical data or expert experience, candidate control actions (such as "gas flow + 600m 3 / h”). Execute candidate actions and monitor the effects (such as furnace temperature change rate, furnace pressure fluctuation). If the verification passes, store the new rule in the database and update the hash table.

[0126] Through self-learning, the knowledge base coverage is improved to adapt to emerging new operating conditions. This also prevents control failures caused by missing rules. Furthermore, through feedback loops, rule weights are corrected, gradually improving control accuracy.

[0127] The comparison results are as follows:

[0128]

[0129] Example 2:

[0130] After a heating furnace adopts a new control solution:

[0131] Response speed: Control instruction generation time is shortened from 50s to 5s;

[0132] Exception handling: Unmatched working conditions are reduced by 70%, and the self-learning module generates 10-15 new rules per month;

[0133] Reduced energy consumption: Annual gas cost savings of approximately 2 million yuan.

[0134] This embodiment uses SQL screening, hash table acceleration and self-learning expansion to achieve high efficiency in rule matching and dynamic optimization of the knowledge base, solving the problems of slow response and low coverage of traditional expert systems, and significantly improving the real-time, adaptability and energy-saving effects of heating furnace control.

[0135] In some embodiments of the present invention, the self-learning module updates the knowledge base in the following manner:

[0136] Collect statistics on historical control effects and calculate the success rate of rule execution;

[0137] If the success rate is lower than the threshold, the rule priority is lowered or marked as invalid;

[0138] If the newly added working condition data conflicts with the existing rules, new rules are generated and assigned initial weights.

[0139] In this embodiment, the system records the control results of each rule after each execution (such as furnace temperature recovery time, furnace pressure fluctuation amplitude, energy consumption change, etc.). If a rule is executed 100 times, and 80 of them make the furnace temperature error ≤10°C, then the success rate is 80%. The threshold is set according to engineering experience (such as a success rate <70% triggers downgrade). Move the rule from a high-priority rule set to a low-priority rule set to reduce the matching probability. Completely exclude the rule to avoid continuing to execute inefficient operations. A rule was downgraded because its success rate was only 60%, and it will only be triggered when no other rules are matched.

[0140] When the newly added working condition data (such as temperature error = -50℃) cannot match the existing rules, it is marked as a conflict. Based on the successful operation of similar working conditions in the past, candidate rules are generated (such as "When the temperature error is ≤ -50℃, the gas flow rate is +1200m 3 / h”). The initial weight is set according to the confidence level of historical data (e.g., initial weight = 0.8). After a short-term trial run, if the control effect meets the standard (e.g., the error recovery speed is improved by 20%), it will be officially stored in the knowledge base.

[0141] The self-learning module eliminates inefficient rules and improves the overall efficiency of the knowledge base. Furthermore, it continuously generates new rules based on operating conditions to adapt to different operating scenarios. This enables continuous optimization of the knowledge base, significantly improving control accuracy, energy efficiency, and system adaptability. Its core value lies in upgrading the traditional static expert system to a dynamic, self-evolving system, making it particularly suitable for complex and changing industrial heating scenarios.

[0142] In some embodiments of the present invention, the commutation period includes:

[0143] Dynamically adjust the burner reversing batches according to the exhaust gas temperature curve, adopt a dispersed reversing method, and only reversing part of the burners each time;

[0144] When the reversing command is triggered, the exhaust fan frequency is adjusted synchronously to offset the furnace pressure shock.

[0145] In this embodiment, distributed commutation involves dividing the burners into batches and switching only a portion of them (e.g., 1 / 3 of the total number of burners) at a time. For example, if the system has 12 burners, 4 burners are commutated at a time, completing the commutation in 3 batches. The commutation frequency is adjusted based on real-time changes in the exhaust gas temperature curve (e.g., temperature fluctuation amplitude and gradient). For example, if the exhaust gas temperature fluctuates by more than ±20°C, the commutation interval is shortened (from 10 minutes to 5 minutes).

[0146] At the moment of burner reversal (e.g., switching from combustion mode to exhaust mode), the system calculates the change in furnace pressure in real time and dynamically adjusts the exhaust fan frequency. For example, if reversal causes the furnace pressure to drop by 50 Pa, the exhaust fan frequency will increase by 3 Hz to compensate for the pressure loss. The actual furnace pressure is fed back by a pressure sensor, and a PID algorithm is used to dynamically adjust the fan frequency to ensure that furnace pressure fluctuations are within ±15 Pa.

[0147] By adopting a decentralized reversing method, the sudden change of instantaneous exhaust volume can be reduced, thereby reducing the fluctuation amplitude of furnace pressure and improving the stability of the system.

[0148] Dynamically adjusting the switching batches based on the exhaust gas temperature curve avoids localized high or low temperature areas, narrows the air-fuel ratio error range, and improves gas utilization. At the same time, dispersed switching makes heat distribution more uniform, reducing the oxidation and burning rate of steel billets during heating.

[0149] Dynamic adjustment of exhaust fan frequency reduces ineffective operation time and fan energy consumption. It also reduces the mechanical impact of concentrated burner reversing, reduces equipment failure rate, and extends maintenance cycle.

[0150] By combining decentralized switching with synchronized exhaust fan regulation, this solution effectively addresses the furnace pressure surges and energy waste associated with traditional switching methods, ensuring efficient and stable operation of the heating furnace. Its core innovation lies in its dynamic response to flue gas temperature changes and its closed-loop control to offset pressure fluctuations, resulting in the combined technical advantages of energy conservation, stability, and longevity.

[0151] In some embodiments of the present invention, the air-fuel ratio correction includes:

[0152] Dynamically adjust the air flow rate based on the deviation between the measured and predicted flue gas oxygen content;

[0153] When the deviation exceeds the preset value, the air-fuel ratio emergency correction rule is triggered to limit the increase in gas flow.

[0154] In this embodiment, by real-time monitoring of flue gas oxygen content deviations, the air flow rate is dynamically adjusted to ensure the air-fuel ratio approaches the theoretical optimal value, reducing gas waste and improving thermal efficiency. This embodiment uses a 5% deviation as an example. When the deviation exceeds 5%, an emergency correction rule is triggered to limit the increase in gas flow rate, avoiding the production of harmful gases such as CO due to incomplete combustion and preventing drastic fluctuations in furnace temperature. By precisely controlling the air-fuel ratio and reducing gas consumption, energy savings can be significantly improved. This also improves the completeness of gas combustion and reduces pollutant generation.

[0155] It's worth noting that gas calorific value analysis, combined with heat balance calculations, can be used to search historical calorific value values ​​to guide the calculation of theoretical air volume at the corresponding calorific value. Specifically, based on calorific value analysis, the air flow rate can be calculated for a specific gas flow rate. Because calorific values ​​can fluctuate and operating conditions can be unstable, for each calorific value, if the gas flow rate is fixed, a corresponding air flow rate can be calculated in the database through heat balance calculations, thus determining the air-fuel ratio. If the air-fuel ratio is not appropriate and deviates from the predicted value, the air flow rate is dynamically adjusted.

[0156] In some embodiments of the present invention, further comprising:

[0157] Integrate fuzzy logic algorithm into the inference engine to fuzzify the rule premise conditions;

[0158] When there is a conflict in rule matching, the membership function is used to weight and generate the final control instruction.

[0159] In this embodiment, the precise input parameters (such as temperature error, furnace pressure) are converted into fuzzy sets (such as "low temperature", "normal", "high temperature"), and the degree to which they belong to each set is quantified by the membership function. According to the membership of the input parameters, the activation strength of each rule is calculated (the minimum value or product of the membership of the premise conditions is taken). The conclusion operation (such as the gas flow adjustment amount) is mapped to a fuzzy output set (such as "small increase", "medium increase", "large increase"). When multiple rules are triggered at the same time, the output fuzzy set is weighted according to the activation strength of each rule to form a comprehensive output. The fuzzy output is converted into a precise control quantity using the center of gravity method, maximum value method, etc. For example, the weighted average value is 640m 3 / h as the final command.

[0160] Fuzzy logic allows input parameters to partially satisfy multiple conditions (e.g., the coexistence of "slightly elevated" and "normal" temperatures), avoiding the "either-or" limitations of traditional rules and covering more transitional operating conditions. For example, when the temperature error reaches a critical value (e.g., ET = M ± 5°C), the system can still generate reasonable instructions rather than relying solely on threshold judgment. By combining the outputs of multiple rules with weighted membership, the control variable changes continuously and smoothly, avoiding the frequent starting and stopping of gas valves caused by traditional rule switching.

[0161] In some embodiments of the present invention, the expert knowledge base supports offline expansion, including:

[0162] Add new rules through manual input and assign initial priorities;

[0163] Import external data sources to generate generalization rules.

[0164] In this embodiment, by allowing engineers to quickly supplement expert experience based on actual working conditions, sudden or special working condition problems (such as equipment abnormalities, new billet heating requirements) are solved. By importing historical data from other heating furnaces or similar scenarios, universal rules are generated to reduce dependence on single scenario experience and improve system adaptability. A combination of manual rules and data-driven rules is adopted to ensure that the knowledge base can reflect expert experience and absorb the common laws of multi-source data to avoid rule rigidity. When a new heating furnace is deployed, the knowledge base can be quickly initialized by importing the rule base of the existing system to shorten the debugging cycle. Manual rules can cover extreme working conditions (such as sensor failures), and external data generalization rules enhance the system's adaptability to unknown working conditions and reduce the risk of control failure.

[0165] refer to Figure 7As shown, the present invention also discloses a heating furnace control system, including a memory 10 and a processor 20, wherein the memory 10 stores a determination program for heating furnace control based on expert control, and the processor 20 runs the determination program for heating furnace control based on expert control to enable the heating furnace control system to execute the above-mentioned heating furnace control method based on expert control.

[0166] The present invention also discloses a computer-readable storage medium, comprising: a determination program for heating furnace control based on expert control is stored on the computer-readable storage medium, and when the determination program for heating furnace control based on expert control is executed by a processor, the above-mentioned heating furnace control method based on expert control is implemented.

[0167] Since the heating furnace control system and the computer-readable storage medium adopt all the technical solutions of the heating furnace control method based on expert control of the above embodiment, they at least have all the beneficial effects brought by the technical solutions of the above embodiment, which will not be repeated here.

[0168] The embodiments of the present invention are described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Various changes can be made within the knowledge of ordinary technicians in the relevant technical field without departing from the scope of the present invention.

Claims

1. A heating furnace control method based on expert control, characterized in that: The following steps are involved: Constructing an expert knowledge base: Using production rules to represent combustion control knowledge, the rule format is IF <precondition> THEN <conclusion operation>. The preconditions include furnace temperature error range, furnace pressure status, flue gas temperature, and reversing cycle parameters. The conclusion operation is gas flow adjustment, air-fuel ratio correction, flue valve opening, or burner reversing cycle setting. Dynamic parameter prediction and optimization: Calculate the predicted temperature value under the current working conditions based on the furnace temperature prediction model. The model is: T m (k+1)=f(T(k),T(k―1),gas(k),gas(k―1),gas(k―2)) Among them, T m To predict the furnace temperature, gas is the gas flow rate, f is the nonlinear function; k is the time; Multi-objective collaborative reasoning: By using a heuristic search matching rule set and combining it with a hash algorithm, the control conclusion can be quickly located. The heuristic function is: e(x)=G(x)―M(x) Where G(x) is the weighted thermocouple measured temperature, and M(x) is the model predicted value; Execution control instructions: Send the optimized air-fuel ratio, reversing cycle and furnace pressure parameters to the PLC, driving the actuator to adjust the gas valve, air valve and exhaust fan frequency; Self-learning and knowledge base update: Dynamically modify rule weights based on historical control effects, add or delete rules, and optimize knowledge base priority sorting.

2. The heating furnace control method based on expert control according to claim 1, characterized in that: The production rules are stored by priority, and the rule set priority is divided according to the furnace temperature error range: When E T <―M, the air-fuel ratio adjustment rule is executed first; When|E T When |≤M, the furnace pressure stabilization rule is executed first; When E T >M, the gas flow increase rule is executed first; Among them, E T is the error value between the model predicted temperature and the actual detected temperature, and M is the preset threshold.

3. The heating furnace control method based on expert control according to claim 1, characterized in that: The heuristic search specifically includes: Filter the initial rule set through SQL statements; A hash table is used to establish the mapping relationship between the premise and the conclusion, and the conclusion is directly located through the LookUp() function; If the match fails, the self-learning module is triggered to generate new rules and store them in the knowledge base.

4. The heating furnace control method based on expert control according to claim 3, characterized in that: The self-learning module updates the knowledge base in the following ways: Collect statistics on historical control effects and calculate the success rate of rule execution; If the success rate is lower than the threshold, the rule priority is lowered or marked as invalid; If the newly added working condition data conflicts with the existing rules, new rules are generated and assigned initial weights.

5. The heating furnace control method based on expert control according to claim 1, characterized in that: The commutation cycle includes: Dynamically adjust the burner reversing batches according to the exhaust gas temperature curve, adopt a dispersed reversing method, and only reversing part of the burners each time; When the reversing command is triggered, the exhaust fan frequency is adjusted synchronously to offset the furnace pressure shock.

6. The heating furnace control method based on expert control according to claim 1, characterized in that: The air-fuel ratio correction includes: Dynamically adjust the air flow rate based on the deviation between the measured and predicted flue gas oxygen content; When the deviation exceeds the preset value, the air-fuel ratio emergency correction rule is triggered to limit the increase in gas flow.

7. The heating furnace control method based on expert control according to claim 1, characterized in that: Also includes: Integrate fuzzy logic algorithm into the inference engine to fuzzify the rule premise conditions; When there is a conflict in rule matching, the membership function is used to weight and generate the final control instruction.

8. The heating furnace control method based on expert control according to claim 1, characterized in that: The expert knowledge base supports offline expansion, including: Add new rules through manual input and assign initial priorities; Import external data sources to generate generalization rules.

9. A heating furnace control system, characterized in that: It includes a memory and a processor, wherein the memory stores a determination program for heating furnace control based on expert control, and the processor runs the determination program for heating furnace control based on expert control to enable the heating furnace control system to execute the heating furnace control method based on expert control as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that include: The computer-readable storage medium stores a determination program for heating furnace control based on expert control. When the determination program for heating furnace control based on expert control is executed by a processor, the heating furnace control method based on expert control according to any one of claims 1 to 8 is implemented.

Images