A Reservoir Dam Monitoring Technology Integrating Improved ABC Algorithm and SVM Algorithm

Abstract

A reservoir dam is a water conservancy project with large investment and high social and economic benefits, which plays an irreplaceable role in flood control, power generation, water storage, and urban water supply. There is a risk of accidents in the process of reservoir dams, so dam monitoring is an important means to achieve the safe operation of reservoirs. In this paper, taking advantage of the high-dimensional and nonlinear characteristics of dam monitoring data samples, the fusion-improved ABC (artificial bee colony) algorithm is introduced, and the SVM (support vector machine) algorithm is used to optimize the penalty factor and kernel function parameters. The test results of the ABC and SVM algorithm are relatively stable, with small fluctuation amplitude, which can continuously monitor water level, pore water pressure, dam deformation, temperature, humidity, vibration, and other indicators is less than 10%, which is significantly lower than the standard ABC algorithm, the standard ANN algorithm, and the standard SVM algorithm. The independence and characteristics of the ABC–SVM algorithm are significantly higher, and the correlation is 0.03, the RMS (root mean square) is 0.2334, which is lower than that of the standard ABC algorithm of 0.09, and the standard ANN algorithm of 0.8. The stability of the results and performance stability are analyzed, which is greater than 90%. The ABC and SVM is used to predict the displacement and deformation law of the reservoir dam.

1. Introduction

As an important facility in water conservancy projects, the dam has the functions of water storage, soil and water conservation, and supporting agricultural irrigation, and plays a supporting role in hydropower generation and waterway transportation. However, during the operation of the dam, it is subjected to pressure of more than a dozen atmospheres, and a small leak can cause devastating damage to surrounding towns and farmland. Therefore, strengthening dam detection and timely early warning are urgent problems to be solved in water conservancy projects [1,2]. At present, dam safety monitoring has the characteristics of long monitoring time and wide range, which will cause great pressure on computers and servers, so some scholars believe that the integration of intelligent algorithms and quantitative analysis algorithms into dam monitoring can reduce the monitoring data of dams and improve the monitoring effect. Some scholars believe that the wide application of dam intelligent detection equipment also increases the complexity of monitoring data and the difficulty of monitoring, although the previous genetic algorithm, the logic analysis algorithm, and the bee colony algorithm can meet the monitoring requirements of dams, the calculation process is complex and occupies a large number of computing resources and cannot be calculated quickly. Combined with water level data, the ABC algorithm can conduct qualitative analysis of massive data, simplify the complexity of data, and provide a basis for subsequent comprehensive calculation. Some scholars believe that the ABC algorithm can simplify the complexity of the calculation, but the accuracy of the calculation results is poor, and the location of dam anomalies cannot be realized, so it needs to be supported by complex intelligent algorithms. The purpose of the genetic algorithm is to explain the iterative relationship between parent and child nodes, and it lacks directional analysis. The bee colony algorithm is a search algorithm that generates nectar sources, and although it can find the location of the anomaly, the direction of the anomaly cannot be determined. There are positive and negative changes in the monitoring of dam pressure, soil and water conservation, and leakage, and it is impossible to make effective predictions based on abnormal analysis alone, so it is necessary to combine algorithms with vector characteristics. Therefore, some scholars have introduced the SVM algorithm into dam data monitoring, which can achieve qualitative and quantitative comprehensive analysis, and make more effective judgments through vector direction. However, the scope of dam monitoring is extensive, and a single monitoring algorithm cannot be realized, so a comprehensive analysis algorithm should be used to integrate the ABC algorithm (artificial bee colony algorithm) and the SVM (support vector machines) algorithm, but the fusion effect is controversial and lacks practical case support. For example, small and medium-sized reservoirs do not have safety monitoring equipment, while large reservoirs do not have safety monitoring facilities, but do not reach the level of fully automated monitoring, so the application of the SVM algorithm is not effective [3,4]. To improve the scope of application of the monitoring effect, the ABC algorithm and the SVM algorithm can be combined to make judgments based on the actual situation.

Dam safety monitoring mainly includes dam deformation monitoring, bottom stress monitoring, dam seepage monitoring, and water flow velocity monitoring. Among them, monitoring indicators, such as deformation and seepage, can be measured by the ABC algorithm, and pressure and water flow velocity can be measured by the SVM algorithm, combined with manual verification and on-site investigation to improve the accuracy of monitoring [5]. In addition, the ABC algorithm analysis of monitoring time and detection season can better solve the problem of a large amount of monitoring data and, at the same time, for bottom monitoring and seepage monitoring, the introduction of the support vector machine as an auxiliary can also improve the effect of monitoring and provide support for unqualitative monitoring. Some scholars believe that the support vector machine is suitable for the detection of dam deformation and bottom stress [6], and the combination of the support vector machine and the ABC algorithm can more clearly reflect the effect of dam deformation. Therefore, some scholars believe that the combination of dam deformation and the support vector machine can elaborate on the operation of the dam and verify the overall condition of the dam through historical hydrological data and test results [7]. Some scholars believe that, in the process of dam monitoring, there will be external interference, so reducing the interference of monitoring data is also a problem that needs to be solved. The pre-processing of monitoring data by the ABC algorithm can extend the validity of the data to a range, so the impact of interfering data can be reduced. The combination of the ABC algorithm and historical data also makes up for the shortcomings of qualitative analysis and improves the effectiveness of analysis results [8]. In summary, the combination of the ABC algorithm and the SVM algorithm can meet the monitoring needs of dams in theory, and realize the preprocessing of massive data, but it lacks practical case support. On this basis, the reservoir dam was studied, the monitoring data for May–June were collected, and the ABC algorithm was used to classify it. Then, the monitoring and analysis were carried out in combination with the local hydrological characteristics, and the SVM algorithm was used to calculate and output the actual results. Finally, compared with the standard SVM and neural network analysis algorithms, the effectiveness of the proposed algorithm is verified. The purpose of this paper is to strengthen the monitoring of reservoir dams, improve the accuracy of monitoring results, and provide support for related research. The specific research process is shown in Figure 1.

According to the data analysis in Figure 1, it can be seen that, in the monitoring process of dams, it is necessary to collect fixed and dynamic data and conduct intelligent analysis to complete the comprehensive judgment of the data. The collected data not only includes meteorological, hydrological, temperature, humidity, and historical data but needs to predict the development law of the data. Therefore, the calculation process is relatively complex. Previous studies only conducted simple data comparisons in the data analysis process and used seasonal databases for prediction. The analysis process is relatively simple, and the comparison data is large. The research results have significant deviations, and the data is complex, making it difficult for manual judgment. Therefore, this paper improves the SVM and ABC algorithms to quantitatively and comprehensively judge historical data and simplify the analysis of classified data. This not only achieves dam data analysis but effectively improves the quantitative and comprehensive judgment of historical data. Moreover, it improves the accuracy and effectiveness of the division, making it easier for manual revision and analysis in later stages. Therefore, the ABC–SVM algorithm proposed in this paper has theoretical feasibility, can make up for the shortcomings of the original single algorithm, and can achieve comprehensive judgment, verification, and effectiveness of indicators for dams. Therefore, the research process has strong practical and theoretical significance.

2. Monitoring and Analysis of Reservoir Dam Deformation

2.1. Description of the Principle of ABC Algorithm

The deformation monitoring of reservoir dam safety monitoring is affected by uncertain factors, such as weather and hydrology, and the dam monitoring data is temporal and needs to be continuously monitored, so it is necessary to introduce an uncertainty coefficient to improve the prediction accuracy of the algorithm [9,10]. The ABC algorithm is a heuristic algorithm that continuously searches dam data to identify information, such as pressure, flow, and leakage, calculates the difference between each index and the early warning standard [11,12], finds the most suitable anomaly, and outputs the final index. In the analysis process of the ABC algorithm, the speed of identifying outliers in dam monitoring is the early warning speed.

(1)

Input of dam monitoring data

Using various monitoring data as analysis conditions and as input data for the model, and conduct post analysis, clarifying the sources of each data when obtaining input data.

The ABC algorithm randomly generates the initial warning index according to Equation (1), then the dam safety standard value $S_n$ and the dam outlier value are 1, the dam warning vector is $f_{i,j}^{warn}$, and its construction process is $f_{i,j}^{strac}$, $f_{i,j}\in \left(1,2,\dots ,d\right)$ is the difference between the early warning standard and the actual test results, and is calculated as shown in Equation (1).

$$f_{i,j}=f_{i,j}^{\min }+monit\left(0,1\right)\cdot \left(f_{i,j}^{warn}-f_{i,j}^{strac}\right)+\Delta \lambda$$

(1)

In Equation (1), $monit\left(0,1\right)$ is the degree of monitoring, $j$ is the dam outlier, and $\Delta \lambda$ is the coefficient of uncertainties. In the process of searching for abnormal data at the dam, within a certain probability range, it is necessary to determine its core function, and if the most dangerous function point is found, leave early warning marks, such as pressure, water flow and displacement, etc., and continue to calculate the surrounding range until all data points have been searched.

(2)

Processing of data acquisition for each monitor

Process the data from the monitor and set an initial threshold to filter whether the obtained data is reasonable. The initial value of the kernel function calculation is the historical data of the dam in the region, such as the previous dam pressure and leakage rate, etc., and the basic value is calculated based on this value, and the maximum value within the range of the kernel function is searched and compared with the early warning value. In general, the range of the calculation of the number of probability kernel functions, the magnitude of the range is determined by the early warning criteria [13,14], and the kernel function is $p_i$, the calculation of which can be described by Equation (2).

$$p_i=fi{t}_i/{\displaystyle \sum _{i=1}^{S_i}fi{t}_i}$$

(2)

In Equation (2), $S_i$ is the $i_{th}$ dam warning suitability $fi{t}_i$ is described, resulting in Equation (3).

$$fi{t}_i=\left\{\begin{array}{l}1/\left(1+va{l}_i\right),if va{l}_i\ge 0\\ 1+abs\left(va{l}_i\right),if va{l}_i<0\end{array}\right.$$

(3)

In Equation (3), $va{l}_i$ is the value of the $i_{th}$ objective function, $abs\left(va{l}_i\right)$ is corresponding to the $i_{th}$ dam warning is represented.

(3)

Perform algorithm analysis on the processed data and output the final result

The dam warning position was $F_i$ not replaced after continuous search, and the dam safety standard value of the location was transformed into an extended index, and the extended index repeated Equation (1) and then regenerated the initial value to calculate the adaptive value of the new dam warning.

However, the ABC algorithm has certain defects, which will lead to the decline of displacement diversity and poor rapid convergence performance when there are more local optimal early warning indicators in the objective function or there is a valley circle near the optimal early warning indicators. Due to the convergence of displacement, the algorithm is easy to jump out of the local optimal early warning index in the later search. However, chaos has the characteristics of randomness and erodibility [15] and can traverse all states according to its laws within the search field, so this paper introduces taboo table and chaos search to improve the ABC algorithm and jumps out of the optimal early warning index of the ABC algorithm through the search probability of chaos theory [16,17]. The main steps of chaos search can be described as follows:

Step 1. Set a random initial vector sequence of D-dimension as $X_i=m_{0,1},m_{0,2},\dots m_{0,D}$, $m_{0,k}\in \left(0,1\right)$ and the difference between the values is small.

Step 2. Based on the initial vector $X_0$, $mix\left(1-X_i\right)$ is a mixture sequence, which is generated by applying the logistic Equation (4).

$$X_i=\mu \cdot mix\left(1-X_i\right)$$

(4)

In Equation (4), the control variable is $\mu$, at the $\mu =4$ time, the traversal process is completed.

Step 3. The dam monitoring process uses a mixing degree series, or the new early warning indicators are Equation (5).

$$V_{ij}=X_{ij}+rand\left(-1,1\right)\left(X_{ij}-X_{kj}\right)$$

(5)

In Equation (5), the number of the early warning indicator is $i$, the result is $j\in \left\{1,2,\dots ,D\right\}$, then $k\in \left\{1,2,\dots ,N\right\}$ is a randomly generated coefficient. Where, $i\ne k$, if the adaptive value of the new warning indicator is greater than the adaptive value of the old warning indicator, the old value will be updated. Otherwise, the number of times the old value fails to be updated is increased by 1.

Step 4. If the maximum number of dam monitoring times has been reached, the optimization process is completed; if not, return to step (2).

The introduction of chaotic search in the search range of dam safety standard value can reduce the probability of falling into the local minimum, and the optimization effect of the function is tested after preliminary optimization by combining with other algorithms, and the optimization effect of the ABC algorithm is better.

2.2. Dam Displacement Prediction Algorithm with the Improved ABC and SVM

As a type of new intelligent optimization algorithm, the ABC algorithm has the problem of falling into local optimum in the practical application process of the ABC algorithm [18,19,20], which solves the contradiction between the accurate search between the newly expanded dam warning and the known dam warning $C$. Therefore, the improved ABC algorithm is introduced on the SVM algorithm, and the penalty factor and kernel parameters of the least squares support vector machine (SVM) are optimized with the improved algorithm $\sigma$. The specific optimization steps can be described as follows.

Step 1. Description of transverse and longitudinal displacement.

The control parameters in the CABC algorithm are initialized, and the displacement size is set to 2N, $N=120$ is the number of extended indicators, $x_{0,k}\in \left(0,1\right)$ is the number of indicators observed in the same dimension, the number of early warning indicators is the limit value, $\lim N$ is the maximum number of cycles, and the initial dimension is $X_0=x_{0,1},x_{0,2},\dots x_{0,k}$, among others.

Step 2. Comprehensive displacement

Based on the initial vector, which is $X_0$, the mixture sequence $X_0$ is generated by using the logistic equation $X_0,X_1,X_2,\cdots X_N$ to $\mu$ to represent the control variables, $\mu =4$ and the process is completely traversal.

Step 3. Set the fitness function of dam monitoring in the ABC algorithm, and calculate the fitness value of each displacement and pressure

Step 4. Extended indicators: New early warning indicators are obtained by using chaotic sequences (2)

If the adaptability of the new warning indicator is greater than the adaptation value of the old warning indicator, it can be recorded as updating the old warning indicator; Otherwise, the number of update failures for the old alert indicator is increased by 1.

Step 5. the possible values of each early warning indicator is $Q_i$, Calculated as Equation (6).

$$Q_i={\displaystyle \frac{f\left(X_i\right)}{{\displaystyle \sum _{n=1}^{N}f\left(X_n\right)}}}$$

(6)

In Equation (6), the number of the early warning indicator is $i$, $j\in \left\{1,2,\dots ,D\right\}$ then $k\in \left\{1,2,\dots ,N\right\}$ is randomly generated. If the adaptive value $i\ne k$ of the new warning indicator is greater than the adaptive value of the old warning indicator, the old value will be updated. Otherwise, the number of times the old value fails to be updated is increased by 1.

Step 6. If the maximum number of dam monitoring times has been reached, the optimization process is completed; If not, return to step (2).

3. Results and Discussion

3.1. Data Analysis for Reservoir Dam Monitoring

Lechang Gorge Water Conservancy Hub is located in Shaoguan, Guangdong Province. The catchment area above the dam site is 4988 km². The flood control storage capacity is 211 million m³, and the total storage capacity is 344 million m³. It is a roller-compacted concrete gravity dam, classified as a level 2 building. The maximum dam height is 83.2 m, the crest length is 256.0 m, and the crest width is 7.0 m. Aerial photos taken by the drone are shown in Figure 2.

In 2019, the safety assessment work for the hub dam was completed. The Lechang Gorge Water Conservancy Hub Dam has been rated as a “Class 1 Dam”, with the project quality level assessed as “Qualified” and the operation management as “Standardized”.

The safety monitoring scope of the Lechang Gorge Water Conservancy Hub project mainly includes the dam project, left and right bank slopes, water diversion system, underground powerhouse, reservoir bank landslide bodies, and associated auxiliary projects. The monitoring items include environmental factors, dam body deformation, uplift pressure, seepage around the dam, seepage flow, anchor stress, steel stress, surrounding rock deformation, concrete strain, and temperature monitoring. Monitoring instruments, such as tension wires, inverted wires, bimetallic markers, static level gauges, joint meters, uplift pressure pipes, piezometers, flow meters, rebar meters, steel plate meters, and strain gauges, have been installed. The data collection for dam safety monitoring is conducted through manual observation and automated observation. Specifically, 302 instruments are installed in the river-blocking dam, 51 instruments on the left and right bank slopes, 83 instruments in the water diversion system, 25 instruments in the diversion tunnel plug, 282 instruments in the underground powerhouse, 71 instruments in the reservoir bank landslide bodies, 62 instruments around the camp area, and 11 instruments in the scour holes on the left and right banks downstream of the river-blocking dam.

The water level in the reservoir is 205.4 m (the warning water level is 200 m). The pore water pressure is 150 kPa. The deformation of the dam body is 0.5 mm in the transverse direction and −0.3 mm in the longitudinal direction. The monitoring temperature is 29 ± 1°, the vibration is 2 Hz, and the vibration amplitude is 0.1 mm. The average annual rainfall is 30 mm. The wind speed is 518 km/h. The dam is regularly inspected once a month, randomly checked every March, and the parameters are recorded. For the data that sticks to the device, it is backed up every two months and uploaded to the cloud. The test time for this study is March to October 2022, and the data are collected at the gravity dam. The total amount of data collection was 50 M, and the data collection time was 30 days. The monitoring data is complex comprehensive data, and the output results are qualitative results, early warning or normal. The monitoring personnel find the corresponding monitoring process by outputting the results. The test location is shown in Figure 3.

We have selected data from 21 monitoring points for our analysis. Among these, some are highlighted in the safety monitoring point distribution map, as shown in Figure 3. The test points are mainly located in the dam body and show symmetrical dispersion results. Among them, the test point is located at the bottom of the dam, and key indicators, such as pore pressure and frequency, are collected. Dam surface test points mainly collect data, such as rainfall and temperature. The data test algorithms are theoretical analysis, manual verification, and comparison with actual results. Among them, the test data is subject to the average value, and the change process of the analyzed value is shown in Figure 4.

According to Figure 4, the dam data collected the test results, as well as the rational distribution of the dam data, and the detection data was calculated through a step-by-step iteration. It is found that, when qualitative data accounts for 40% and quantitative data accounts for 60% of the test data, the data rationality is higher. Among them, the qualitative data mainly includes historical data, meteorological data (rainfall, wind), etc., and the quantitative data mainly includes water level, pore water pressure, dam deformation, temperature, humidity, vibration, etc. The results, as shown in Figure 4, suggest that the quantitative data monitoring of 21 points should be the mainstay, and the qualitative data should be supplemented. Moreover, it is necessary to carry out multiple monitoring at the monitoring points to improve the effectiveness of the monitoring results. The total amount of data collection was 50 M, and the data collection time was 30 days.

3.2. Accuracy of Reservoir Dam Monitoring

The optimal data results after 15 training sessions of the fusion improvement optimization algorithm were obtained, the variables in the regression algorithm were selected based on the stepwise regression processing, and the decision variables were randomly generated based on the result information in a small range, so the search efficiency and accuracy were effectively improved. The convergence speed of all comparison algorithms needs to be less than 160 steps. After 80 iterations, the convergence speed of the ABC algorithm shows a downward trend. Observing the values in the figure, it can be seen that the ABC–SVM algorithm has high convergence accuracy and better performance than the other algorithms, as shown in

Table 1. Convergence and improvement of the calculation accuracy of different algorithm.

Relevant Index	ABC and SVM Algorithm	Standard ABC Algorithm	Standard SVM Algorithm	Standard ANN Algorithm
Water level [m (%)]	193.94 (94.42)	173.40 (84.42)	143.2 (83.12)	132.41 (81.12)
Pore water pressure [kPa (%)]	144.23 (96.15)	129.83 (86.55)	119.23 (85.32)	107.24 (80.33)
Deformation of dam body [mm (%)]	0.49 (98.62)	0.044 (88.73)	0.021 (68.23)	0.004 (78.11)
Temperature [°C (%)]	26.75 (92.24)	23.09 (82.41)	20.22 (80.31)	19.32 (78.31)
Humidity [% (%)]	42.53 (94.52)	38.04 (84.54)	36.33 (79.32)	36.33 (79.32)
Vibration [m (%)]	9.10 (91.00)	8.10 (81.01)	7.89 (78.22)	7.02 (72.13)

.

According to the data analysis in Table 1, it can be seen that the accuracy of the ABC and SVM algorithm during dynamic analysis is greater than 90%, significantly better than the ABC algorithm, the ANN algorithm, and the standard SVM algorithm. In order to verify the sustainability of the results of this study, the test results were continuously analyzed, and the specific results are shown in Figure 5.

As can be seen from Figure 5, the test results of the ABC and SVM algorithm are relatively stable, although the fluctuation frequency is high, but the fluctuation amplitude is small, mainly because the frequent fluctuation is to compare the monitoring data with the historical data, and then adjust the test parameters of the quantitative test indicators, such as water level, pore water pressure, dam deformation, temperature, humidity, vibration, etc., to improve the test effect. Relatively speaking, the ABC algorithm has a large change range and a low adjustment frequency, mainly because the algorithm does not find the abnormality of the test results, which indicates that the ABC algorithm cannot identify the abnormality of the monitoring data in time and cannot find the abnormal data of dam changes. It can be understood that the ABC and SVM algorithm can accurately identify the abnormality of the monitoring data and achieve the purpose of monitoring; the main reason for this is to classify qualitative and quantitative data, quickly eliminate normal monitoring data, and not only save monitoring hardware resources but improve the monitoring effect.

3.3. Accuracy of Results in Reservoir Dam Monitoring

To effectively monitor dam data, it is necessary to perform significance analysis on qualitative and quantitative data and compare the results with the standard ABC algorithm. Among them, the calculation results are mainly the correlation between the algorithms and the accuracy of the algorithm results, as shown in

Table 2. The accuracy of indicator monitoring results for different algorithms [%].

Monitoring Indicators	Standard ABC Algorithm	ABC and SVM Algorithm	Standard SVM Algorithm	Standard ANN Algorithm
Water level	75.36	95.12	85.25	78.63
Pore water pressure	82.62	92.48	75.14	74.35
Deformation of the dam Body	72.03	95.69	74.02	82.36
Temperature	76.95	85.89	65.24	72.32
Humidity	85.36	91.07	75.42	71.96
Vibration	92.63	97.79	72.32	68.75

.

The analysis in Table 2 shows that the ABC and SVM algorithm has significant advantages in analyzing water level, air pressure, dam deformation temperature, and qualitative indicators. Its overall calculation accuracy is greater than 85%, significantly better than the standard ABC algorithm. At the same time, comparing it with the commonly used ANN algorithm in dam monitoring, it can also be found that the algorithms proposed in this paper is superior to the ANN algorithm. The reason for the comparisons with the ANN algorithm is that, during the analysis process, the data for dam monitoring is obtained using the ANN algorithm and should be measured through the built-in internal analysis system to conduct a statistical analysis on the results of the test indicators and determine whether there is independence in the results. Whether the test results have significant characteristic indicators, the specific results are summarized in

Table 3. The Statistical analysis of dam monitoring indicators.

The Index of Statistical	ABC and SVM Algorithm	Standard ABC Algorithm	Standard SVM Algorithm
Independence of results (%)	0.88	0.81	0.72
Result characteristics (%)	0.92	0.43	0.68
Correlation of results (%)	0.03	0.09	0.42
Degree of conformity with random sampling results (%)	0.94	0.72	0.71
Degree of agreement between theoretical prediction and actual test (%)	0.89	0.73	0.63

.

It can be seen that the independence and characteristics of the ABC–SVM algorithm are significantly higher than those of the standard ABC algorithm, and the correlation is 0.03, which is lower than 0.09 for the standard ABC algorithm, indicating that the independence and characteristics of the calculation index are better. Through random spot check and analysis of prediction results, it is found that the coincidence degree of random spot check results is 94%, which is significantly higher than the 72% results for the standard ABC algorithm. By comparing the theoretical prediction with the actual test results, it is found that the ABC–SVM algorithm is higher than the standard ABC algorithm. On the whole, the independence and conformity of the calculation results are good, which meets the monitoring needs of the dam. It also proves that the ABC and SVM algorithm can effectively process comprehensive data and achieve the goal of effective monitoring. By continuously monitoring the test results, the final output results are obtained, as shown in Figure 6.

It can be seen from Figure 6 that the error change of the ABC and SVM algorithm is small, and the accuracy improves greatly under the adjustment of data structure and test times, which indicates that the results of dam multi-point monitoring are accurate, and the conclusion is consistent with the results shown in Figure 5. In the process of dam monitoring data, at 18 iterations, the calculation results of the ABC algorithm change, while the calculation process of the ABC and SVM algorithm does not change. By comparing the error changes of the two algorithms, it is also confirmed that the errors of the ABC and SVM algorithm fluctuate slightly, and the calculation process is stable. The main reason is that the ABC algorithm simplifies the qualitative data, while the SVM algorithm mines the quantitative data, and analyzes the amplitude, water level, dam body, and pore water pressure indexes combined with the attributes of test points to realize the complexity analysis of the key indexes of the dam, so the calculation results are accurate and meet the actual monitoring requirements.

3.4. Predictability of Reservoir Dam Monitoring

Although dam indicators are analyzed independently, the logical relationship between indicators needs to be mined. Temperature, humidity, rainfall, and wind direction are also auxiliary indicators for dam monitoring. Therefore, it is necessary to strengthen the analysis of the relationship between the auxiliary indexes of dams and the main measurement indexes. Among them, the relationship parameter between the main index and the auxiliary index is the fitting value, so the fitting value between the indexes should be calculated. In this paper, the predicted value of the dam displacement of the reservoir is compared with the actual value by integrating the ABC and SVM algorithm and the standard ABC algorithm, and the specific comparison results are shown in Figure 7.

Observing the data in Figure 6, the fitting value between the auxiliary index and the main index is good, which shows that the auxiliary index can provide reference for the continuous monitoring of the dam and verify the calculation results of the main index. When analyzing dam fitting values, there is a problem of calculation deviation, so RMS error verification is required. The specific results are shown in

Table 4. Actual prediction and analysis results of the ABC–SVM algorithm.

Statistical Indicators	ABC–SVM Algorithm	Standard ABC Algorithm	Standard SVM Algorithm
RMS error	0.2334	0.2564	0.2623
Sum of the remaining squares	71.2	75.2	76.35

.

The data in Table 4 shows that the fitting error of the ABC and SVM algorithm is small, and the proposed algorithm has higher prediction accuracy and stability due to the standard ABC algorithm. The main indicators and auxiliary indicators reflect the operation law of the dam, monitor the deformation of the dam, the compressive resistance of the dam body, and the characteristics of water storage, and predict the monitoring results. Moreover, the ABC and SVM algorithm can monitor the monitoring data and have a high ability to process complex data, which can realize the continuous monitoring of the dam, provide comprehensive information for the dam management personnel, and effectively improve the safety prediction of the dam, as shown in

Table 5. Comparison of the robustness of different algorithms.

Statistical Indicators	ABC–SVM Algorithm	Standard ABC Algorithm	Standard SVM Algorithm	Standard ANN Algorithm
Performance deviations	82.3	85.6	77.33	76.63
Stability deviations	86.2	35.2	35.85	37.64

.

From the data analysis in Table 5, it can be seen that the performance deviation and stability deviation of the ABC–SVM algorithm proposed in this paper are good, and the overall deviation results are good, both greater than 80%. Relatively speaking, the standard ABC algorithm and the standard SVM algorithm have good bias, but the stability bias is poor. The overall performance advantage of the ABC and SVM algorithm is 82.3%, with a stability of 86.2%, which is significantly better than the standard ABC algorithm’s 85.6% and 35.2%, the standard SVM algorithm’s 77.33% and 35.85%, and the standard ANN algorithm’s 76.63% and 37.64%. Therefore, the results of the ABC–SVM algorithm are significantly better than those of the other algorithms in terms of bias testing. In order to verify the adaptability of the research algorithm in this paper, the stability test of the results should be carried out in different environments, and the specific results are shown in

Table 6. Comparison of stability and universality under different constraints.

Index	Restraint	ABC and SVM Algorithm	Standard ABC Algorithm	Standard SVM Algorithm	Standard ANN Algorithm	X2	p
stability	Daytime	33.33	22.36	18.63	10.32	18.000	0.324
	Nighttime	22.34	18.63	20.65	16.35
	Summertime	28.96	25.32	16.35	10.62
	Wintertime	46.95	33.69	44.37	38.62
	Spring- and falltime	25.75	22.36	18.63	17.32
universality	Daytime	32.61	17.32	29.66	28.96	15.000	0.241
	Nighttime	29.65	28.63	20.77	19.75
	Summertime	13.85	10.66	12.68	11.36
	Wintertime	45.96	43.39	36.89	30.88
	Spring- and falltime	31.72	17.32	29.66	27.61

.

From the data analysis in Table 6, it can be seen that the application adaptability of the ABC–SVM algorithm proposed in this paper is higher than that of the standard ABC algorithm and the standard SVM algorithm, and there are significant differences (p > 0.05 x2 > 15). Among them, its stability and generality are better during the daytime period. The daytime stability of the ABC and SVM algorithm is 33.33%, and the nighttime stability is 22.34%, which is significantly better than the ABC algorithm’s 22.36% and 18.63%, as well as the SVM algorithm’s 18.63% and 20.65%, respectively. At the same time, it is also better than the ANN algorithm’s 10.32% and 16.35%. At the same time, in the general positive aspects of winter, spring, and autumn seasons, the ABC and SVM algorithm is significantly better than the ABC algorithm at 43.39% and 17.32%, and higher than the SVM algorithm at 36.89% and 29.66% and the ANN algorithm at 30.88% and 27.61%. The results show that the algorithm proposed in this paper is suitable for other related monitoring of dams, and can better use hydrological data, historical data, and pore pressure data to meet the multi-faceted safety monitoring requirements of dams. Therefore, this paper proposes that the algorithm is more adaptable and suitable for the depth monitoring of dam safety.

3.5. Discussion of Study Results

3.5.1. Error Analysis of Dam Monitoring Results

By testing and verifying the bus analysis results using the ABC and SVM algorithm and outputting the relationship between each result through a phase diagram, the visualization of the overall analysis effect is improved. The specific results are shown in Figure 7.

As shown in the analysis in Figure 8, the output results of the ABC algorithm, the SVM algorithm, and the ANN algorithm are both smaller than those of the ABC and SVM algorithm, indicating significant differences between the four algorithms. Further verification of the effectiveness of the analysis in this article is mainly due to the classification of dam monitoring data using the ABC algorithm, and the comprehensive judgment of quantitative data and analysis content using the SVM algorithm. The relationship between the two is output through the comprehensive analysis results of quantitative and qualitative analysis, and effective warning of the danger level of the dam is given. This not only improves the efficiency of its analysis but enhances its accuracy. The ABC algorithm can effectively reduce the amount of input data in the early stage and improve the effectiveness of later analysis by classifying dam monitoring data, such as pressure voltage, current, etc., and removing redundant data.

3.5.2. The Trend of Data Changes in Monitoring Results

In order to better analyze the monitoring results and achieve an effective comparison between the various results, a three-point chart analysis can be conducted on the results, and the three-point chart can be planned to achieve a trend analysis of the three-point chart. The specific results are shown in Figure 9.

The results in Figure 9 draw scatter plots for the ABC, SVM, and ANN algorithms, and predict the trend of changes in the analysis results of the three algorithms. The results of the analysis show that the trend of the three algorithms is almost consistent, indicating that, in the process of analysis, the ABC and SVM algorithm, and its analysis performance, is significantly better than the ABC, SVM, and ANN algorithms. Moreover, all three algorithms have a lack of consistency in the process of monitoring dam data, making it impossible to comprehensively analyze quantitative information. So, the analysis results showed a straight line without significant fluctuations, indicating that the ABC and SVM algorithm proposed in this paper can effectively monitor dam data. The main reason for this is that the ABC algorithm can simplify the complexity of various data devices in dam data monitoring, establish logical relationships between devices, and reduce the difficulty of data processing in the early stage. Through the SVM algorithms in the later stage, the processing results are deeply analyzed to form effective data correlation analysis, thereby improving the accuracy of data output results.

4. Conclusions

Dams are the key to hydropower hubs and require continuous and effective monitoring to ensure their stable operation. However, factors such as water level, humidity, rainfall, and wind direction can affect the accuracy of dam monitoring. Therefore, this paper proposes the ABC and SVM algorithm to realize the comprehensive analysis of qualitative and quantitative data and verify their effect on dam monitoring. The results show that the test results of the ABC and SVM algorithm are relatively stable, with small fluctuation amplitude, which can continuously monitor water level, pore water pressure, dam deformation, temperature, humidity, vibration, and other indicators, with a monitoring error at less than 10%, which is significantly lower than the standard ABC algorithm, the standard SVM algorithm, and the standard ANN algorithm. Moreover, in terms of the recognition effect of the calculated indicators, the independence and characteristics of the ABC joint SVM algorithm are significantly higher than those of the standard ABC algorithm, and the correlation is 0.03, which is lower than that of the standard ABC algorithm of 0.09, and the consistency of the random sampling results is 94%, which can effectively predict the safety of dams. At the same time, the stability of the results and performance stability of the algorithm in this paper are analyzed, which is greater than 90%, and the fitting error between the indicators is small, due to the standard ABC algorithm. Therefore, the ABC and SVM algorithm can not only continuously monitor the dam but predict the possible future hazards and monitor the depth of the dam, which can meet the requirements of all aspects of monitoring. There are also some limitations in this study, mainly using historical data and humidity data to filter subjective analysis, which affects the integrity of the data, and the k-means algorithm will be introduced for data screening in the future to improve the calculation effect.